Pipes Feed Preview: Matt Ryan, PhD & Feed not found & Hackread – Cybersecurity News, Data Breaches, AI and More & The Hacker News & Seriously Risky Business - Risky.Biz & Feed not found & CyberScoop & Feed not found & SecurityWeek & The Record from Recorded Future News & darkreading & Feed not found

  1. FBI Investigating Apparent Hack of Its Key National Criminal Justice Database, AP Source Says

    Sun, 22 Nov 2026 00:53:02 -0000

    <p>The FBI is investigating an apparent hack of a national law enforcement database that it maintains.</p> <p>The post <a href="https://www.securityweek.com/fbi-investigating-apparent-hack-of-its-key-national-criminal-justice-database-ap-source-says/">FBI Investigating Apparent Hack of Its Key National Criminal Justice Database, AP Source Says</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  2. President Trump Taps Kirsten Davies as National Cyber Director

    Sat, 21 Nov 2026 13:00:00 -0000

    <p>White House says the battle-tested cyber leader will drive an America First cyber strategy.</p> <p>The post <a href="https://www.securityweek.com/president-trump-taps-kirsten-davies-as-national-cyber-director/">President Trump Taps Kirsten Davies as National Cyber Director</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  3. Hackers Using Email Bombing Before Posing as IT Support on Microsoft Teams

    Sat, 21 Nov 2026 12:54:31 -0000

    Fortra warns of an email bombing campaign where hackers flood work inboxes, then pose as IT support on&#8230;
  4. Amazon: China-Linked Hackers’ AI Agents Exploited Medical Software Zero-Day

    Sat, 21 Nov 2026 12:08:50 -0000

    <p>A threat actor used autonomous Cairn AI agents to discover and exploit a critical MONAI Label vulnerability.</p> <p>The post <a href="https://www.securityweek.com/amazon-china-linked-hackers-ai-agents-exploited-medical-software-zero-day/">Amazon: China-Linked Hackers&#8217; AI Agents Exploited Medical Software Zero-Day</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  5. A Single Malicious Webpage Let Researchers Hijack Meta Muse and Access Private Chats

    Sat, 21 Nov 2026 11:53:07 -0000

    A malicious webpage could extract a Meta Muse user's private conversations and stored memories from the assistant without triggering any approval prompt, researchers have shown. The same attack could be extended to reach the credentials Muse stores for the user's connected accounts. The work was a laboratory demonstration, not an attack on real users, and it stayed inside the individual user's
  6. GnuPG Warns Critical GPGME Flaw Could Expose App Data via Malicious OpenPGP Keys

    Sat, 21 Nov 2026 11:05:53 -0000

    A malicious OpenPGP key with an unusually long user ID can trigger a critical memory bug in GPGME, the standard library that programs use to work with GnuPG. The GnuPG project warned yesterday that the flaw could be used to steal data from affected applications and has released fixed versions. The bug, tracked as CVE-2026-129975, affects all software that uses GPGME before version 2.3.0,
  7. Attackers' AI Agents Found a Zero-Day in Medical-Imaging Software and Gained Root Access

    Sat, 21 Nov 2026 11:03:17 -0000

    Suspected China-linked hackers used autonomous AI agents to find a previously unknown flaw in MONAI Label, an open-source medical imaging server, and gain root access to a university server, Amazon Threat Intelligence said this week. The flaw allows an attacker with no account to write files outside the server's storage folder, and it affects MONAI Label 0.8.5 and earlier in the default setup,
  8. Fake GTA VI Streaming Client Installs Data-Stealing Malware on Windows PCs

    Sat, 21 Nov 2026 08:57:36 -0000

    Bitdefender warns of an active GTA VI streaming scam targeting Windows gamers with malware that can steal stored passwords and cryptocurrency wallets.
  9. Hackers Exploit Critical VMware, Cisco, and Apache Flaws as Espionage Attacks Emerge

    Sat, 21 Nov 2026 08:49:13 -0000

    The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Friday added four flaws impacting Apache Roller, Cisco Expressway, SAP NetWeaver, and Broadcom VMware vCenter to its Known Exploited Vulnerabilities (KEV), citing evidence of active exploitation. The vulnerabilities are listed below - CVE-2026-129448 (CVSS score: 9.8) - A deserialization of untrusted data vulnerability in
  10. West Virginia Legislature hit with cyber incident affecting IT systems

    Fri, 20 Nov 2026 21:34:36 -0000

    A cybersecurity incident that began Thursday was still ongoing Friday in the West Virginia Legislature. A spokesperson said the disruption did not spread to other government entities.
  11. China's 'Armory Castle' Lays Siege to VMware vCenter Zero-Day

    Fri, 20 Nov 2026 21:03:27 -0000

    During the last wave of major VMware vCenter security updates, a China-nexus threat group likely already had its next zero-day up its sleeve.
  12. US companies’ use of Chinese open weight AI draws rebuke from House chairmen

    Fri, 20 Nov 2026 20:25:00 -0000

    An investigation led by the House Homeland Security Committee focused on the dangers of U.S. companies using lower-cost open weight AI models for certain projects.
  13. Container Store restores systems after cyberattack

    Fri, 20 Nov 2026 19:32:00 -0000

    The firm’s parent company, Neighborhood Intelligence, said in a statement that The Container Store has “restored its website and systems following a cybersecurity incident this past weekend.”
  14. In Security, Words Matter as Much as Tools, Maybe More

    Fri, 20 Nov 2026 18:35:40 -0000

    From talking cloud security to using AI, cyber leaders are finding that clear, disciplined communication is now a front-line defense capability across the enterprise.
  15. WhatsApp Patches 3 Security Flaws, Including Android Bug Suspected of Use in Scams

    Fri, 20 Nov 2026 17:57:20 -0000

    Meta-owned WhatsApp has addressed security vulnerabilities, including one impacting its Android app that it suspects may have been exploited in connection with scam messages. The vulnerability, tracked as CVE-2026-23871 (CVSS score: 3.5), has been described as a case of incomplete validation of external ad-reply URLs in WhatsApp for Android and WhatsApp Business for Android before 2.26.17.4.
  16. Amazon says Chinese hackers used Cairn AI tool to discover apparent zero-day during attack on university

    Fri, 20 Nov 2026 17:30:00 -0000

    A China-linked threat actor used the open source AI penetration testing tool Cairn to persistently attack two universities, and in one case, possibly discover a zero-day bug, researchers said.
  17. What We Missed: Nvidia CEO Scandal Rocks AI Industry

    Fri, 20 Nov 2026 17:04:04 -0000

    In this video, Dark Reading editors discuss some of the news they didn't get a chance to cover, from Jensen Huang's resignation from Nvidia to Microsoft's recent cloud outage.
  18. Bazuka Android Malware Uses Fake Screen Overlays to Hide Credential Theft and Device Control

    Fri, 20 Nov 2026 16:58:16 -0000

    Cybersecurity researchers have disclosed details of a previously unknown Android malware framework called Bazuka that offers large language model (LLM) integration for victim management and control at scale. Dutch mobile security company ThreatFabric said it discovered the threat in late October 2026 during an investigation into an NFC-relay banking Trojan. "Bazuka is a full malware framework,
  19. WhatsApp Fixes Flaw Uncovered While Disrupting Scam Campaign

    Fri, 20 Nov 2026 16:01:02 -0000

    <p>Tracked as CVE-2026-23871, the vulnerability was found by WhatsApp’s security team, and it may have been used to make scam messages more credible.</p> <p>The post <a href="https://www.securityweek.com/whatsapp-fixes-flaw-uncovered-while-disrupting-scam-campaign/">WhatsApp Fixes Flaw Uncovered While Disrupting Scam Campaign</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  20. In Other News: Norwegian Energy Firm Cyberattacks, Supercomputer Hack, New Nightmare Eclipse Exploit

    Fri, 20 Nov 2026 14:56:23 -0000

    <p>Noteworthy stories that might have slipped under the radar:  EU adds NoName057(16) to cyber sanctions list, malware targets CACs left in personal computers, Gemini agents targeted Epoch AI.</p> <p>The post <a href="https://www.securityweek.com/in-other-news-norwegian-energy-firm-cyberattacks-supercomputer-hack-new-nightmare-eclipse-exploit/">In Other News: Norwegian Energy Firm Cyberattacks, Supercomputer Hack, New Nightmare Eclipse Exploit</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  21. Linux KVM Flaw CVE-2026-99312 Lets Guest VMs Access Host Memory on AMD Hosts

    Fri, 20 Nov 2026 12:20:00 -0000

    A flaw in KVM, the hypervisor built into the Linux kernel, let researchers read about 127 GB of a cloud server's memory from inside a virtual machine of their own on DigitalOcean. On a shared server, that memory can include other customers' virtual machines. The flaw, tracked as CVE-2026-99312, affects only hosts with AMD processors and nested virtualization enabled. Fixed kernels shipped
  22. SBC Medical Group Data Breach Impacts Over 2 Million

    Fri, 20 Nov 2026 11:36:51 -0000

    <p>Threat actors accessed databases containing customer information associated with clinics in Japan.</p> <p>The post <a href="https://www.securityweek.com/sbc-medical-group-data-breach-impacts-over-2-million/">SBC Medical Group Data Breach Impacts Over 2 Million</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  23. European Commission Confirms AI-Assisted Cyberattack Disrupted Services

    Fri, 20 Nov 2026 10:51:42 -0000

    <p>The Commission took certain services offline to contain the incident, but it has no evidence of a data breach.</p> <p>The post <a href="https://www.securityweek.com/european-commission-confirms-ai-assisted-cyberattack-disrupted-services/">European Commission Confirms AI-Assisted Cyberattack Disrupted Services</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  24. SonicWall Warns Gen 6 Firewalls Won't Get Security Patches, Even for Critical Flaws

    Fri, 20 Nov 2026 10:38:22 -0000

    SonicWall has urged all customers still running Generation 6 firewalls to retire and replace them immediately and stop exposing the devices on the internet edge, citing security risks fueled by AI-assisted attacks. "Attackers now use automation and AI to find and exploit unsupported edge devices faster than ever," SonicWall said. "We expect internet-facing Gen 6 firewalls to be actively
  25. Ruby Patches Ruby2Shell Vulnerability That Could Lead to RCE

    Fri, 20 Nov 2026 10:34:10 -0000

    <p>Tracked as CVE-2026-132825, the bug is a heap-based buffer overflow that could lead to remote code execution (RCE).</p> <p>The post <a href="https://www.securityweek.com/ruby-patches-ruby2shell-vulnerability-that-could-lead-to-rce/">Ruby Patches Ruby2Shell Vulnerability That Could Lead to RCE</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  26. Hackers Exploit Critical Cisco Expressway Flaw, Breach Four Devices and Reach Internal Network

    Fri, 20 Nov 2026 10:32:50 -0000

    Threat actors have been observed exploiting a recently disclosed critical security flaw in Cisco Expressway to breach a Russian organization shortly after the public release of a proof-of-concept (PoC) exploit. The activity involves the exploitation of CVE-2026-76466 (CVSS score: 9.8), a vulnerability that could allow an unauthenticated, remote attacker with network access to a Cisco Expressway
  27. China-Linked Hackers Exploit Critical VMware vCenter Zero-Day to Gain Admin Access and Steal Data

    Fri, 20 Nov 2026 09:38:15 -0000

    The suspected China-nexus cyber espionage group known as Armory Castle (formerly UNC5221) has been attributed to the zero-day exploitation of a critical security flaw in VMware vCenter Server in early September 2026. The vulnerability, tracked as CVE-2026-123890 (CVSS score: 9.8), is a case of authentication bypass in the VMware Directory Service that could be exploited to gain unauthorized
  28. N-able Blames Service Disruption on Cyberattack

    Fri, 20 Nov 2026 09:08:35 -0000

    <p>The incident affected Mail Assure and SpamExperts Hosted Cloud and could have led to credential theft.</p> <p>The post <a href="https://www.securityweek.com/n-able-blames-service-disruption-on-cyberattack/">N-able Blames Service Disruption on Cyberattack</a> appeared first on <a href="https://www.securityweek.com">SecurityWeek</a>.</p>
  29. ROMULUS ClickFix Attacks Deliver RAT That Can Spy on Users and Control PCs

    Fri, 20 Nov 2026 07:11:39 -0000

    Threat actors have been observed using the ROMULUS ClickFix malware delivery platform to deliver a previously undocumented remote access trojan (RAT) that shares similarities with CNCMachineRMS. "ROMULUS likely operates as a delivery service, distributing CNCMachineRMS and Legion Loader through separate chains while retaining ClickFix and signed binary sideloading techniques," Blackpoint
  30. Sean Cairncross out as national cyber director

    Fri, 20 Nov 2026 01:07:07 -0000

    <p>During his tenure, he was involved in key administration decisions on AI security and the protection of critical infrastructure.</p> <p>The post <a href="https://cyberscoop.com/sean-cairncross-out-national-cyber-director/">Sean Cairncross out as national cyber director</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Sean Cairncross is leaving his post as national cyber director, CyberScoop has learned.</p> <p class="wp-block-paragraph">Cairncross led the Office of the National Cyber Director, which coordinates cybersecurity policy and strategy across the federal government. Though he had experience in Republican politics and national security, he lacked a technical or cyber background. His office nevertheless became involved in key administration decisions on AI security and the protection of critical infrastructure.</p> <p class="wp-block-paragraph">A source close to the White House told CyberScoop that President Donald Trump fully supported firing Cairncross from his position as he lacked the strong cyber credentials to perform the job.</p> <p class="wp-block-paragraph">Department of Defense Chief Information Officer <a href="https://defensescoop.com/tag/kirsten-davies/">Kirsten Davies</a> will move into an acting role in the White House office, a transition that comes amid staff departures and debate over how the government should manage artificial intelligence risks. </p> <p class="wp-block-paragraph">CyberScoop has reached out to the Pentagon for comment. </p> <p class="wp-block-paragraph"><a href="https://x.com/_KirstenDavies_/status/2123595758465089690">In a post on X</a>, Davies thanked Trump and White House Chief of Staff Stephen Miller for the appointment. <br><br>&#8220;Our nation stands in the crossroad of analog, digital, and quantum technologies, our American way of living increasingly interconnected. Our opportunity is to dominate the cyber landscape for generations to come …but only if we modernize and drive resilience into the foundations our great nation runs on, as well as build cyber talent for our future,&#8221; she wrote. &#8220;Under President Trump’s leadership we will have a more secure, resilient; and supercharged technological future.&#8221;</p> <p class="wp-block-paragraph">Cairncross met with technology company leaders and helped shape a June executive order that asked leading AI firms to submit new models for voluntary government testing before public release. He also was assigned to co-lead a working group tracking a nonbinding AI security commitment signed by major labs.</p> <p class="wp-block-paragraph">Cairncross advocated for collaboration between government and industry, arguing that <a href="https://cyberscoop.com/trump-ai-executive-order-open-source-strategy-sean-cairncross/">security measures should not slow</a> the development of AI. The administration has sought to deploy AI in critical infrastructure as a way to strengthen defenses, while lawmakers and some executives have called for stronger government guardrails. President Donald Trump has rejected new regulation.</p> <p class="wp-block-paragraph">The office’s work extended beyond AI policy. It launched a program offering cybersecurity resources to <a href="https://cyberscoop.com/watershed-250-texas-water-cybersecurity-pilot/">water utilities in Texas</a> after attacks on water systems and a six-month pilot in South Carolina aimed at rural hospitals. <a href="https://cyberscoop.com/oncd-rural-hospital-cybersecurity-pilot-project-lifeline-250/">Project Lifeline 250</a> connects companies, state agencies and local governments to help hospitals strengthen network defenses, keep critical care operating after an incident and replace vulnerable or aging equipment.</p> <p class="wp-block-paragraph">The projects reflect the challenges facing sectors with limited technology resources. Rural hospitals, like water utilities, often have fewer resources than other critical infrastructure operators. Federal officials have also cited healthcare as a major target for ransomware and other cyberattacks.</p> <p class="wp-block-paragraph">The leadership change at ONCD follows a period of staff turnover. Cairncross’ policy chief and senior adviser Thomas Lind <a href="https://www.citybiz.co/article/913606/openai-taps-former-trump-adviser-thomas-lind-to-lead-cyber-and-strategic-risk-team/">joined OpenAI in October</a>. Other senior staff members departed after the administration released its AI executive order, including an adviser involved in early policy negotiations.</p> <p class="wp-block-paragraph">The turnover comes as the administration attempts to fill other cyber leadership posts. Michael LoSapio, Trump’s nominee to lead the Cybersecurity and Infrastructure Security Agency, faced <a href="https://cyberscoop.com/michael-losapio-cisa-confirmation-hearing/">a contentious confirmation hearing Thursday</a>. CISA helps protect federal civilian networks and critical infrastructure.</p> <p class="wp-block-paragraph">Cairncross’s departure also comes after Trump administration officials blamed <a href="https://cyberscoop.com/tech-outages-ai-vulnerability-patching/">a spate of recent tech outages</a> on a race to close vulnerabilities uncovered by artificial intelligence.</p> <p>The post <a href="https://cyberscoop.com/sean-cairncross-out-national-cyber-director/">Sean Cairncross out as national cyber director</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  31. Qwen Incidents Show Self-Hosted AI Models Need Stronger Steering

    Thu, 19 Nov 2026 22:40:03 -0000

    Companies deploying AI agents have the illusion of control, but several incidents demonstrate that businesses need to be ready to enforce directions.
  32. Navigating Vulnerability Hysteria in the AI Era

    Thu, 19 Nov 2026 19:46:44 -0000

    This installment of the Reporters' Notebook video series examines the fallout from recent high-severity flaw disclosures, exploring how fears of AI-assisted exploits trigger widespread panic, operational disruptions, and growing warning fatigue.
  33. ThreatsDay: GPT-6.2 Jailbreak Flaws, BubbleLoader PNG Malware, $500M Ransom Demands + 17 More Stories

    Thu, 19 Nov 2026 18:47:58 -0000

    The easiest way to fool someone is to give them something they already expect. A job offer. A message about a service outage. A familiar app. Nothing looks strange at first. And that's the problem. This week's stories show how ordinary things are being used to steal data, spread malware, and get people to hand over access. There's another shift worth watching. AI is making some attacks easier
  34. Dems call CISA nominee ‘disqualified’ after nomination hearing

    Thu, 19 Nov 2026 17:11:34 -0000

    <p>Michael LoSapio, the Palantir executive and Trump administration nominee for CISA, sparred with Democrats on the committee over who won the 2020 election, Jan. 6 and more.</p> <p>The post <a href="https://cyberscoop.com/michael-losapio-cisa-confirmation-hearing/">Dems call CISA nominee ‘disqualified’ after nomination hearing</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Michael LoSapio, the Palantir executive and Trump administration nominee for Director of the Cybersecurity and Infrastructure Security Agency, struck a combative tone with Democrats on the Homeland Security and Government Affairs Committee and took sides in several partisan disputes that are likely to complicate his path to confirmation.</p> <p class="wp-block-paragraph">LoSapio’s hearing lasted less than an hour, far less time than Congress has spent scrutinizing previous CISA nominees, as he was considered alongside a large tranche of other Trump administration nominees that Republicans are attempting to move through Congress during the lame duck session.</p> <p class="wp-block-paragraph">CISA has <a href="https://cyberscoop.com/nick-andersen-cisa-acting-director-nomination/">lacked a confirmed director</a> for the first two years of the Trump administration, and policymakers have <a href="https://cyberscoop.com/cisa-nick-andersen-cyber-resilience-outages/">fresh concerns</a> about cyber insecurity in U.S. critical infrastructure, artificial intelligence and other threats. Despite that, LoSapio and senators on the committee spent little time discussing the specifics of his vision for the federal government’s primary defensive cybersecurity agency.</p> <p class="wp-block-paragraph">When LoSapio was asked about cybersecurity issues, like how he intended to boost cybersecurity resilience of the nation in the wake of last week’s widespread outages, he said organizations should be more proactive implementing baseline protections around inventory, evaluation&nbsp; and patching internet-connected devices.</p> <p class="wp-block-paragraph">“It’s really important to understand the landscape of the systems you have out in the wild,” said LoSapio. “The things that you operate day-to-day, the systems that you rely on, the systems that are ancillary, and do an extensive evaluation on their vulnerabilities, whether they’ve been patched or not, how they’ve been architected and their defensive capabilities, both things we need to add and the things that exist today, as to how that maps to the threats we’re facing.”</p> <p class="wp-block-paragraph">At another point he stated that there was “more we can do across the nation” to improve security, but did not articulate beyond that. He did not commit to hiring more staff, only to evaluate the agency’s resources once confirmed.</p> <p class="wp-block-paragraph">But LoSapio also waded into several partisan debates and made comments that led Sen. Richard Blumenthal, D-Conn., to state that he had “disqualified” himself from the role of CISA director by demonstrating a lack of independence from the White House.</p> <p class="wp-block-paragraph">At one point, LoSapio said he didn’t view the Jan. 6, 2021 riots led by Trump supporters on Capitol Hill as an “attack,” doubling down on the assertion when pressed repeatedly by Blumenthal. After the hearing, the social media account for Senate Homeland Security Democrats posted the exchange on X, calling on the Senate to reject LoSapio’s nomination.</p> <p class="wp-block-paragraph">“Anyone who can so easily rewrite history in service to Trump should not be leading CISA, which has critical election security obligations,” the <a href="https://x.com/HomelandDems/status/2123445773819945312">post</a> read.</p> <p class="wp-block-paragraph">In another exchange, Sen. Bernie Moreno, R-Ohio, asked what the impact would be on CISA if “Democrats” did not fund the agency during the lame duck session. LoSapio answered it would be “catastrophic” and likened its impact on “ordinary Americans” to calling 9-1-1 and having no one pick up. While past shutdowns have hindered agency operations, CISA and other DHS component agencies would continue to operate under emergency appropriations during a shutdown.</p> <p class="wp-block-paragraph">Later, when&nbsp;Sen. Elissa Slotkin, D-Mich., asked about the effects of thousands of firings and hundreds of millions of dollars in cuts over the past two years, as well as the White House budget proposal that would further cut hundreds of millions next year, LoSapio said he could not assess them because he was an outsider coming to the agency.</p> <p class="wp-block-paragraph">The late nomination of LoSapio gives him little time between now and January, when a less-friendly, Democratic-majority Congress takes office, to lobby support from the lame duck Senate. His nomination will likely compete for limited floor time as the Trump administration has put forward dozens of federal nominees since the 2026 election.</p> <p class="wp-block-paragraph">Following the hearing, Punchbowl News <a href="https://x.com/Dareasmunhoz/status/2123482316769206363" target="_blank" rel="noopener">reported</a> that retiring Sen. Thom Tillis, R-N.C. announced he would place a hold on LoSapio’s nomination.</p> <p>The post <a href="https://cyberscoop.com/michael-losapio-cisa-confirmation-hearing/">Dems call CISA nominee ‘disqualified’ after nomination hearing</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  35. phpAfterShell Explained: How Hackers Could Turn a PHP Session Flaw Into Code Execution

    Thu, 19 Nov 2026 16:59:56 -0000

    PHP's default file-based session storage can be made to return objects an attacker chose, by cutting off a session write at the right moment. When combined with a second weakness, such as a suitable gadget chain in the app, those objects can allow the attacker to run code on the server. Today's PHP releases fix the flaw, called phpAfterShell. The flaw is tracked as CVE-2026-126714 and rated
  36. PHP Fixes 7 Critical Flaws, Including Bugs That Could Allow Remote Code Execution

    Thu, 19 Nov 2026 15:18:07 -0000

    The PHP Foundation has released security updates to address multiple critical security flaws that could be exploited to achieve remote code execution on affected deployments, including CVE-2026-115560. The release of the patch follows an advance notice in which the project maintainers alerted users to multiple critical vulnerabilities affecting all supported PHP versions under specific
  37. Drupal Patches Critical Installer Flaw That Could Give Unauthenticated Attackers Control of Sites

    Thu, 19 Nov 2026 14:33:06 -0000

    Under specific conditions, a critical vulnerability in Drupal core can allow an attacker without login credentials to use the built-in installer to attempt to reinstall a site. That can lead to a site takeover or remote code execution, and the project rates the vulnerable configurations as uncommon. Fixed releases for all supported branches shipped yesterday. The Drupal Security Team disclosed
  38. Why Holiday Security Can't Wait for Black Friday

    Thu, 19 Nov 2026 14:00:00 -0000

    Bot-driven fraud and holiday social engineering attacks stretch across many weeks, straining security teams and small businesses alike.
  39. F5 Patches Two NGINX HTTP/2 Vulnerabilities, One Dating Back to 2015

    Thu, 19 Nov 2026 12:02:10 -0000

    Two NGINX HTTP/2 vulnerabilities can let attackers run code without logging in. F5 has released fixes for affected versions of NGINX Open Source and NGINX Plus.
  40. Everything Is a Web Application, Including Your Firewall

    Thu, 19 Nov 2026 11:52:23 -0000

    One of the early targets of PHP2Shell exploitation was the firewall. On October 28, Palo Alto Networks confirmed that attackers were exploiting the two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, against PAN-OS management interfaces exposed to the internet. Both flaws let anyone execute code on the web server without authentication. The disclosure itself went a little wrong.
  41. Hackers Breach N-able Cloud Server, Disrupting Email Security Management Across 3 Regions

    Thu, 19 Nov 2026 09:31:26 -0000

    N-able on Wednesday disclosed that unknown threat actors gained unauthorized access to one of the servers in the Mail Assure and Spam Experts Hosted Cloud, causing it to disable management interfaces and APIs. "We have taken steps to mitigate the threat and are continuing to harden our services to reduce the risk of recurrence," N-able said. "The investigation is ongoing, but at this time we
  42. Africa Retakes Top Spot as Most Attacked Region, Led by Nigeria Surge

    Thu, 19 Nov 2026 08:00:00 -0000

    Nigeria faced more than double the cyber threats of the previous year, with Angola and Kenya also in the top 5 most-attacked countries globally.
  43. CISA Warns of 4 Exploited Flaws Linked to Ransomware, Cryptomining, and LLMjacking Attacks

    Thu, 19 Nov 2026 08:00:00 -0000

    The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Wednesday added four vulnerabilities impacting Synology DiskStation Manager (DSM), BerriAI LiteLLM, Apache Axis, and Vanderbilt University Research Electronic Data Capture (REDCap) to its Known Exploited Vulnerabilities (KEV) catalog, citing evidence of active exploitation. The vulnerabilities are listed below -
  44. PHP Warns of Critical Security Flaws Affecting All Supported Versions, Fixes Coming November 19

    Thu, 19 Nov 2026 04:33:46 -0000

    The maintainers of the PHP programming language on Wednesday issued an advisory announcing an upcoming security release that addresses critical flaws. The release is scheduled for November 19, 2026, between 2 and 2:30 p.m. UTC. "These releases are part of PHP's regular four-week release cycle and will address security vulnerabilities affecting all supported PHP versions," the maintainers said.
  45. Srsly Risky Biz: Phew! Some Regulatory Relief for AI Hacks

    Thu, 19 Nov 2026 03:48:49 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://thinkst.com/"><em><u>Thinkst</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/2026-11-19-xan-griffin-eA2t5EvcxU4-unsplash.jpg" alt="Srsly Risky Biz: Phew! Some Regulatory Relief for AI Hacks"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://thinkst.com/"><em><u>Thinkst</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB191.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB191/&quot;&gt;Srsly Risky Biz: Regulatory relief for AI hacks is a good first step&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 17:51 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Regulatory relief for AI hacks is a good first step&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id></h2><figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-222ba9dd-af02-4845-af0d-77867181cb33.jpeg" class="kg-image" alt="Srsly Risky Biz: Phew! Some Regulatory Relief for AI Hacks" loading="lazy" width="2000" height="1335" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/11/data-src-image-222ba9dd-af02-4845-af0d-77867181cb33.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/11/data-src-image-222ba9dd-af02-4845-af0d-77867181cb33.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/11/data-src-image-222ba9dd-af02-4845-af0d-77867181cb33.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-222ba9dd-af02-4845-af0d-77867181cb33.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@xangriffin?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Xan Griffin</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/person-standing-on-rock-raising-both-hands-eA2t5EvcxU4?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>The Trump administration&apos;s AI task force will encourage regulatory agencies to go easy on organisations that fall victim to AI-powered cyber attacks. Last week the task force <a href="https://www.cybersecuritydive.com/news/ai-white-house-task-force-cybersecurity-regulation/874679/"><u>said in a statement</u></a> that AI has changed the risk environment, and therefore &quot;prior standards for taking regulatory action in response to cyber incidents may be outdated&quot;.&#xa0;</p><p>The idea here is to encourage good-faith, voluntary reporting of vulnerabilities and incidents because a wave of strict enforcement actions in the current environment could have a &quot;chilling effect on reporting and information sharing&quot;. The task force described such reporting as &quot;critical to our broad risk mitigation efforts&quot;. The suggested regulatory leniency would last for at least 12 months, depending upon how things pan out.&#xa0;</p><p>This makes sense, at least in the short-term. It would be perverse to punish victims for having security settings that were appropriate last year, but may not be today.&#xa0;</p><p>AI has changed the equilibrium between attackers and defenders, so policymakers are right to adjust standards and expectations to fit the new reality of cheap, easy and fast AI-driven attacks.&#xa0;</p><p>Rather than just giving victims a short-term, get-out-of-jail-free card to deal with the problem, however, we&apos;d like to see more forward-looking policy that is aimed at encouraging organisations to adapt to the new normal.&#xa0;&#xa0;</p><p>The call for leniency, for example, could be expanded to allow organisations greater leeway to experiment with defensive AI efforts. If we want to find our way back to the acceptable equilibrium between attackers and defenders, organisations need some freedom to experiment with AI-enhanced defence. Take AI red teaming, for example. If a test were to go awry despite genuine efforts to be careful, a little government tolerance might be needed.</p><p>Governments should also encourage organisations to raise the priority of cyber security and devote more management attention and even resources to it.&#xa0;</p><p>Finally, governments need to flesh out their practical, AI-specific guidance. We are aware of <em>some</em> high-level advice, a joint Five Eyes cyber security agency <a href="https://www.ncsc.gov.uk/sites/default/files/2026-06/Five-Eyes-cyber-security-agencies-statement-ai-shift.pdf"><u>statement from June</u></a> and a <a href="https://www.cyber.gov.au/sites/default/files/2026-07/Defending%20against%20AI-enabled%20cyber%20attacks%20-%20Guidance%20for%20government%2C%20critical%20infrastructure%20and%20large%20enterprises.pdf"><u>slightly more detailed</u></a> ASD document from July. But &quot;one and done&quot; blog posts from a couple of agencies isn&#x2019;t enough. We&apos;d hope to see these statements released by more governments and they should be living documents with advice that changes in real time with the technology.</p><p>Regulatory leniency will buy a small amount of breathing room, but ultimately it&apos;s a temporary stopgap. Governments should be encouraging sensible experimentation, identifying&#xa0; approaches that work, and then prodding organisations to adopt them.&#xa0;</p><h2 id="americas-rules-vs-chinas-money-fight">America&apos;s Rules vs China&apos;s Money: Fight!&#xa0;</h2><p>On Monday, the Commerce Department&apos;s Bureau of Industry and Security (BIS) <a href="https://www.bis.gov/press-release/bis-issues-rule-addressing-remote-access-american-super-intelligence-technology"><u>published new AI and semiconductor export regulations</u></a> that tighten controls on who can buy AI chips and what they can be used for. The plan is to stop China&apos;s AI labs from being able to acquire Nvidia chips or rent compute from Southeast Asian data centres.</p><p>It&apos;s a nice start, but this won&apos;t stop Chinese AI firms using Nvidia&apos;s chips without a lot more work.</p><p>America&apos;s overall intent here is to maintain its lead in AI by preventing China&apos;s firms from accessing advanced AI chips. The regulation aims to close several commonly exploited loopholes in the old rules.</p><p>For a start the new regulations explicitly make it harder to smuggle chips to China by requiring the US government to review export licenses to Singapore, Malaysia, Thailand, Indonesia, and Brazil. <a href="https://x.com/ChrisRMcGuire/status/2122340547578761513?a=m622zgThh5PPfNv6_LI&amp;s=20"><u>These countries</u></a> are all <a href="https://epoch.ai/data/ai-chip-imports?view=table&amp;tab=by_country&amp;values=quarterly&amp;designer=Nvidia"><u>large importers of Nvidia chips</u></a>, and according to BIS, &quot;present a higher-than-normal risk of diversion&quot; to the PRC.&#xa0;</p><p><a href="https://www.wsj.com/tech/jensen-huang-backdoor-ai-data-centers-a5e4499d"><u>This week</u></a>, for example, <em>The Wall Street Journal</em> reported that Singapore-based cloud company <a href="https://en.wikipedia.org/wiki/Megaspeed"><u>Megaspeed</u></a> arranged to send nearly USD&#x24;12 billion worth of Nvidia chips to China-based companies over a 15 month period from March 2024.&#xa0;&#xa0;&#xa0;&#xa0;</p><p>The new regulation also cracks down on firms from China simply renting access to advanced AI chips hosted in Southeast Asian data centres. This was legal, and there <a href="https://www.cnbc.com/2026/08/19/china-ai-nvidia-chips-us-export-controls.html"><u>are several</u></a> reports <a href="https://carnegieendowment.org/research/2026/05/the-geopolitical-debates-over-controlling-cloud-compute"><u>indicating this</u></a> is a <a href="https://www.chinatalk.media/p/how-many-chips-does-china-have"><u>common practice</u></a>. Senior White House science policy advisor Michael Kratsios <a href="https://x.com/mkratsios47/status/2079933645888880708"><u>claimed Chinese firm Moonshot AI trained</u></a> its <a href="https://news.risky.biz/srsly-risky-biz-knives-are-out-for-open-weight-ai-models/"><u>Kimi K3 model</u></a> in a data centre in Thailand.&#xa0;</p><p>Under the new BIS regulation, foreign data centres will be required to ensure customers in China and arms-embargoed countries cannot access remote compute resources above certain limits.</p><p>It&apos;s great that the new regulation addresses these loopholes, but we expect it will merely make them a bit harder to squeeze through. Closing them properly will require a lot more work.&#xa0;</p><p>The fundamental problem is that the BIS&apos;s regulations are asking companies in Southeast Asia to choose between America&apos;s rules or China&apos;s money. There&apos;s an awful lot of yuan available for companies willing to skirt the BIS&apos;s requirements, when it comes to both remote access and the physical movement of chips.&#xa0;</p><p>Chip smuggling will be a tough nut to crack. In March of this year, the Department of Justice charged the co-founder of American company Super Micro Computer and two Taiwanese co-conspirators with diverting USD&#x24;2.5 billion of servers containing Nvidia chips to China. <a href="https://www.justice.gov/opa/pr/three-charged-conspiring-unlawfully-divert-cutting-edge-us-artificial-intelligence"><u>The DoJ alleged</u></a> that the trio falsified documents and repackaged servers into unmarked boxes to ship them abroad. Prosecutors said they used dummy servers to deceive auditors and that surveillance camera footage showed them removing serial number stickers with hair dryers.</p><p>There&apos;s money in this. Even a small margin on billions of dollars worth of servers adds up to a significant payday and simply tightening the rules won&apos;t stop people from sidestepping them.&#xa0;</p><p>When it comes to remote access, the BIS rule says that data centre operators &quot;will be expected to perform robust due diligence on their customers&quot;. Without strong oversight we suspect that &quot;robust&quot; will be a subjective term that quickly devolves to &quot;bare minimum&quot; without strict audits.</p><p>And here&apos;s the rub. The BIS <a href="https://media.bis.gov/about-bis/bis-leadership-and-offices/OEA/export-control-office-program"><u>only has eleven</u></a> export control officers worldwide. The sole export control officer based in Singapore is also responsible for chip control across Malaysia and Thailand. That&apos;s a lot of territory for one person, even if they are receiving support from other agencies. This year BIS has <a href="https://www.strtrade.com/trade-news-resources/str-trade-report/april/bis-looks-to-significantly-expand-enforcement-in-fy-2027"><u>asked Congress for additional funding</u></a> to bolster its enforcement capabilities, but even if that is granted it will still take time to spin up its activities.</p><p>Preventing China from getting access to advanced AI chips won&apos;t be achieved by simply publishing better regulations. Government agencies need to follow up with on-the-ground enforcement that ensures that the rules are being followed. China&apos;s companies will always be looking for new workarounds, so these rules may need to be updated over time.&#xa0;</p><p>The real headscratcher here is that, for <a href="https://www.nytimes.com/2026/11/13/opinion/jensen-huang-nvidia-howard-lutnick-commerce.html"><u>reasons we don&apos;t understand</u></a>, America has done nothing to enforce chip controls for the last 18 months or so. As a result, China&apos;s AI sector has had a field day.&#xa0;</p><p>These firms have enduring reasons to maintain access to advanced American AI chips, at least for now. The US government needs to respond with a similar lasting effort, rather than sitting back and hoping that 132 pages of regulation will do.&#xa0;</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/Q-bqAPC4etY?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Regulatory relief for AI hacks is a good first step"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Endgame keeps kicking goals:</strong> <a href="https://www.europol.europa.eu/media-press/newsroom/news/64-million-victim-datasets-secured-in-international-operation-against-malware-networks"><u>Europol announced</u></a> that coordinated action by law enforcement agencies and private sector partners resulted in the disruption of three malware networks, Gootloader, CastleRAT and StealC. These are all cybercrime-as-a-service offerings and are used by a variety of criminals to enable ransomware and data theft.&#xa0;&#xa0;</li><li><strong>US freezes &#x24;1.2 billion from Binance hack: </strong>The US Department of Justice <a href="https://www.justice.gov/opa/pr/justice-department-announces-actions-disrupt-north-koreas-massive-theft-cryptocurrency"><u>announced</u></a> that it had frozen the funds and also seized infrastructure used by the North Korean FaceStealer social engineering scheme. It <a href="https://www.fbi.gov/investigate/cyber/facestealer-cryptocurrency-theft-scheme"><u>says FaceStealer</u></a> is &quot;a highly effective social engineering scheme that uses imposter versions of popular video conferencing platforms&quot; and targets companies and individuals in the cryptocurrency industry.&#xa0;</li><li><strong>Lockdown mode for Microsoft Edge:</strong> <a href="https://microsoftedge.github.io/edgevr/posts/Lockdown-Mode/"><u>Microsoft has posted about</u></a> improving the security of the Edge browser including by hardening by rewriting core parts of the browser in the memory-safe Rust language, improving the delivery of security updates and working on what they call Edge Lockdown mode. It goes further than Edge&apos;s Enhanced Security Mode, which disables just-in-time javascript compilation, and also disables other browser features that are also used in attacks.&#xa0;&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Catalin Cimpanu talks with Thinkst CTO Marco Slaviero about the state of the defensive AI cyber space and the company&apos;s current agentic canaries, as well as the need for more experimentation.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI151.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI151/&quot;&gt;Thinkst on the defensive AI cyber space&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 17:30 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Thinkst on the defensive AI cyber space&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss why AI means that mercenary spyware is a dead industry. The pair discuss why governments would rather buy the cow than the milk.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN190.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN190/&quot;&gt;Between Two Nerds: The spyware market after AI&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 33:34 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: The spyware market after AI&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/umv4OuwjOCQ?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: The spyware market after AI"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Hackers hijack Optimizely to deliver malware on 178 popular sites:</strong></p><p>For about 16 hours on Monday, November 16, the Optimizely A/B testing platform delivered malicious code on at least 178 popular websites across the globe.</p><p>According to an<a href="https://status.optimizely.com/incidents/ftw0fh9wx43z"><u> Optimizely statement</u></a>, the hacker exploited a vulnerability in its &quot;Experimentation&quot; platform to add custom code to certain customers.</p><p>According to technical write-ups of what happened&#x2014;via<a href="https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/"><u> Report-URI</u></a>,<a href="https://www.stepsecurity.io/blog/optimjack-optimizely-hosted-scripts-crypto-clipper-card-skimmer?ref=metacurity.com"><u> Step Security</u></a>, and<a href="https://amirulcyber.com/2026/11/16/optimjack-optimizely-skimmer/"><u> Amirul Iman</u></a>&#x2014;the code went through different versions across the day.</p><p>[<a href="https://news.risky.biz/risky-bulletin-hackers-hijack-optimizely-to-deliver-malware-on-178-popular-sites/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Hackers accessed Barracuda&apos;s cloud: </strong>A threat actor has gained access to the cloud environment of security firm Barracuda in a security breach that took place earlier this month and lasted for almost two days.</p><p>In a<a href="https://blog.barracuda.com/2026/11/11/anatomy-of-a-machine-speed-intrusion"><u> post-mortem</u></a>, the company says it traced back the entry point to a cloud access key it issued to a third-party AI service provider and which it forgot to revoke after it finished working with the platform.</p><p>The attacker used the key to access its cloud environment and then deployed the ARTEX AI penetration testing framework to escalate its access across its internal network and other cloud environments.</p><p>[<a href="https://news.risky.biz/risky-bulletin-hackers-accessed-barracudas-cloud/"><u>more</u></a> on <em>Risky Bulletin]</em></p><p><strong>Chinese APT uses SS7, Diameter attacks to spy on targets:</strong> A newly discovered Chinese cyber-espionage group is hacking telcos across Asia to exploit the SS7 and Diameter protocols and spy on carefully selected targets.</p><p>The new SilentAir group has been active for at least two years, since early 2025, according to telco security firm<a href="https://www.enea.com/insights/the-enemy-within-tracking-liminal-panda-and-silentair-through-mobile-networks/"><u> ENEA</u></a>, which has been tracking its operations.</p><p>The group is hacking telco infrastructure and planting backdoors that run malicious SS7 and Diameter commands designed to collect device data, track a subscriber&apos;s location, and even intercept calls and SMS traffic.</p><p>[<a href="https://news.risky.biz/risky-bulletin-chinese-apt-uses-ss7-diameter-attacks-to-spy-on-targets/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  46. Russia's Cold Relic APT Turns One Cloud Breach Into Another

    Wed, 18 Nov 2026 21:59:46 -0000

    The state-sponsored threat group is using data stolen from one victim to target others via convincing impersonations and social engineering attacks.
  47. Consensus mounts that tech outages were fueled by patching of AI-discovered vulnerabilities

    Wed, 18 Nov 2026 20:56:55 -0000

    <p>But cyberattacks aren’t directly to blame, Trump administration officials said.</p> <p>The post <a href="https://cyberscoop.com/tech-outages-ai-vulnerability-patching/">Consensus mounts that tech outages were fueled by patching of AI-discovered vulnerabilities</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Cyberattacks almost certainly aren’t responsible for the procession of <a href="https://cyberscoop.com/cisa-nick-andersen-cyber-resilience-outages/">major tech outages</a> over the past week, Trump administration officials say  — but there’s a growing sense that patching fueled by a race to close vulnerabilities uncovered by artificial intelligence was a contributing factor.</p> <p class="wp-block-paragraph">The top Democrats on the House and Senate intelligence committees are among those observing that the rise in vulnerability discoveries resulting from AI are at least partially to blame for tech service interruptions like those ranging from the aviation ground stop to incidents with major communications and banking systems, rather than the direct result of hacking.</p> <p class="wp-block-paragraph">“What appears to be happening instead is a consequence of how quickly artificial intelligence is changing the cybersecurity landscape,” Sen. Mark Warner, D-Va., said Tuesday. “AI is making it possible to identify vulnerabilities across enormous numbers of systems at unprecedented speed, creating enormous pressure on companies and government agencies to patch those vulnerabilities immediately, often all at once, putting tremendous strain on systems that weren’t built to handle that pace of change.”</p> <p class="wp-block-paragraph">And Rep. Jim Himes, D-Conn., said Wednesday that the trend is likely to continue.</p> <p class="wp-block-paragraph">“It is possible that the increased pace of vulnerability discovery and patching could cause more frequent outages, but the breakneck pace of patching we have seen in recent months is vital to shore up the security of all IT infrastructure,” he said. “Disruptions in networks, whether through internal errors or outside attack[s], are a fact of life, but we can do much more to be prepared and resilient.”</p> <p class="wp-block-paragraph">At a House Homeland Security Committee hearing Wednesday, Trump administration witnesses repeatedly cited patching, rather than hacking, as a cause of the outages, although they didn’t specify that the patches were a result of AI unearthing vulnerabilities.</p> <p class="wp-block-paragraph">“We are in an environment where the rate of patching, the rate of upgrading of systems, is increasing,” said Jay Clayton, director of national intelligence. “That is a result of the advancements in compute, and in applications.”</p> <p class="wp-block-paragraph">But lawmakers were more explicit. The Washington Sun reported that Wednesday that during Hill briefings, industry and government officials cited patching in response to AI-discovered vulnerabilities as <a href="https://cyberscoop.com/cisa-nick-andersen-cyber-resilience-outages/">the cause</a>. Rep. James Walkinshaw, D-Va., responded at Wednesday’s hearing to Clayton testifying that the incidents were “unrelated” to one another.</p> <p class="wp-block-paragraph">“The recent tech outages are not coincidences, and they are connected to cybersecurity,” Walkinshaw said. As cyberattacks are “exploding,” everyone is “frantically patching systems.”</p> <p class="wp-block-paragraph">Walkinshaw noted that as attacks have been on the rise, the Cybersecurity and Infrastructure Security Agency has had its personnel and budget dramatically cut.</p> <p class="wp-block-paragraph">Department of Homeland Security Secretary Markwayne Mullin said that despite the cuts, CISA is “able to meet our mission right now because of new technology that has come in,” but he didn’t say what technology he meant.</p> <p class="wp-block-paragraph">Mullin also said he didn’t have any update for lawmakers about the pace of CISA hiring, as committee members from both sides of the aisle pressed him to increase personnel at the agency.</p> <p class="wp-block-paragraph">Clayton and Mullin, as well as FBI Deputy Director Christopher Raia, repeatedly said that so far, there remains no evidence that malicious cyberattacks had anything to do with the outages. They also said that there was no sign that cyber adversaries were seizing on the time necessary to implement patching to conduct attacks.</p> <p>The post <a href="https://cyberscoop.com/tech-outages-ai-vulnerability-patching/">Consensus mounts that tech outages were fueled by patching of AI-discovered vulnerabilities</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  48. Strike Graph Launches Integration AI to Automate Continuous Compliance Evidence Using Any Existing Tool

    Wed, 18 Nov 2026 20:45:49 -0000

    New agentic AI solution leverages APIs to collect compliance evidence from systems and applies it to over 30 frameworks with no duplicate work.
  49. AttackIQ Research Reveals How Iranian Cyber Groups Hand Off Network Access to Enable Destructive Attacks

    Wed, 18 Nov 2026 20:38:42 -0000

    Analysis of 20 Iranian state-sponsored threat groups details how espionage, ransomware, and hacktivism intersect in state-backed cyber campaigns.
  50. Senate Republicans advance FCC nominee, slate of cyber bills 

    Wed, 18 Nov 2026 19:16:11 -0000

    <p>Confirmation would give Republicans a 3-1 majority on the FCC. Meanwhile, the fates of multiple bipartisan cyber bills are in limbo after Democrats no-showed the markup hearing.</p> <p>The post <a href="https://cyberscoop.com/danielle-thumann-severs-fcc-nomination-tech-bills/">Senate Republicans advance FCC nominee, slate of cyber bills </a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Republicans in the Senate Commerce Committee advanced the nomination of Danielle Thumann Severs to serve as a commissioner at the Federal Communications Commission.</p> <p class="wp-block-paragraph">Alongside Severs’ nomination, the GOP majority also voted favorably on a slew of tech and cybersecurity bills. But Democrats &#8211; who will wield the majority gavel in January – boycotted the hearing, throwing the fate of the bills into question, even as many had bipartisan sponsorship.</p> <p class="wp-block-paragraph">If confirmed, Thumann Severs would give Republicans and chair Brendan Carr a 3-1 majority on the commission, with Biden administration holdover Anna Gomez the only remaining Democrat. The Trump administration has not moved to fire Gomez as it did two Democratic commissioners on the Federal Trade Commission last year. But he also broke with precedent by declining to name a Democratic-appointee alongside Thumann Severs to round out the five-member commission.</p> <p class="wp-block-paragraph">In a hearing room that sat half-empty on the Democratic side, committee chair Sen. Ted Cruz, R-Texas defended the decision to move forward with Thumann Severs’ nomination after resounding losses in the midterm elections.</p> <p class="wp-block-paragraph">“What it should tell us is that the Democrats have zero intention of processing any of the president’s nominees,” Cruz said. “We are heading into two years of total destruction, total war, partisan warfare is what the Democrats are promising us.”</p> <p class="wp-block-paragraph">He argued President Donald Trump is not legally required to nominate FCC commissioners in pairs, and is merely to elect no more than three members of the same party. The nomination of Thumann Severs, he claimed, is “consistent with that requirement.”</p> <p class="wp-block-paragraph">He also suggested that President Trump had demonstrated a desire for the commission to remain bipartisan by declining to fire Gomez after the Supreme Court upheld his right to do so to <a href="https://cyberscoop.com/trump-moves-to-fire-democratic-ftc-commissioners/">FTC commissioners last year</a>.</p> <p class="wp-block-paragraph">CyberScoop has reached out to Democrats on the Senate Commerce Committee for comment. On Tuesday committee member Sen. Ben Ray Luján, D-N.M., <a href="https://www.lujan.senate.gov/newsroom/press-releases/politico-lujan-criticizes-fcc-nominees-candor-calls-for-further-review/">accused</a> Thumann Severs of not being forthcoming about her role in support of Carr&#8217;s media bias investigations.</p> <p class="wp-block-paragraph">Republicans on the committee also advanced multiple tech and cybersecurity-related bills.</p> <p class="wp-block-paragraph"><a href="https://www.congress.gov/bill/119th-congress/senate-bill/69">The COLLUDE Act</a>, sponsored by Sen. Eric Schmitt, R-Mo., would create a private right to sue social media companies that make content moderation decisions based on government requests. Specifically, the bill would strip liability protections under Section 230 of the Communications Decency Act for social media companies, opening them to private lawsuits from individuals if their account or post bans</p> <p class="wp-block-paragraph">As Missouri Attorney General, Schmitt led a lawsuit over the Biden administration’s information sharing with social media companies around COVID-19 and election disinformation. That case eventually reached the Supreme Court, where the majority concluded that the communications were non-coercive and that social media companies have their own, constitutionally-protected incentives for moderating speech on their platforms.</p> <p class="wp-block-paragraph">Under the Trump administration and Carr, the FCC has gone considerably further than the Biden administration in attempting to police political speech over the public airwaves, opening investigations into broadcast media organizations that have aired interviews with Democratic opponents and at times likening their perceived disagreement with the federal government as evidence of broadcasting undue ideological bias. Thumann Severs <a href="https://www.lujan.senate.gov/newsroom/press-releases/politico-lujan-criticizes-fcc-nominees-candor-calls-for-further-review/">served as Carr&#8217;s senior counsel</a> prior to being nominated as FCC commissioner. </p> <p class="wp-block-paragraph">When moving forward Thumann Severs’ nomination earlier in the meeting, Cruz and other Republicans gave only verbal commitments to hold the FCC accountable for its mission to protect free speech.</p> <p class="wp-block-paragraph">Another bill, the <a href="https://www.congress.gov/bill/119th-congress/senate-bill/5508">Telecommunications Cybersecurity and Resilience Act</a>, sponsored by Cruz and Sen. Mark Warner, D-Va., would establish a telecommunications cybersecurity working group at the Department of Commerce that would be charged with developing industry best practices around cybersecurity and develop criteria for independent third-party cybersecurity certifications and assessors.</p> <p class="wp-block-paragraph">The bill comes at the end of the first congressional session since U.S. policymakers were notified that a Chinese hacking group called Salt Typhoon had systematically infiltrated U.S. telecommunications infrastructure and spied on high level U.S. officials, including then-incoming Vice President J.D. Vance.</p> <p class="wp-block-paragraph">But <a href="https://cyberscoop.com/salt-typhoon-senate-commerce-hearing-fcc-telecom-cybersecurity/">both Congress</a> and <a href="https://cyberscoop.com/fcc-cybersecurity-vote-china-hack-telecoms-anna-gomez/">the FCC</a> opted against new legislation or regulation to improve the status quo, instead touting voluntary partnerships that they claim have led to material improvements in cybersecurity across the sector since the Salt Typhoon hacks. Earlier this year, the FCC rescinded new regulatory requirements that were put in place during the final months of the Biden administration.</p> <p class="wp-block-paragraph">The <a href="https://www.congress.gov/bill/119th-congress/senate-bill/3301/cosponsors">Chip EQUIP Act</a>, introduced by Sen. Mark Kelly, D-Ariz., and Marsha Blackburn, R-Tenn.,  would require companies that receive federal funding under the 2022 CHIPS Act to sign a ten year agreement prohibiting the money from being used to procure, install or use of semiconductor manufacturing equipment produced by companies owned or controlled by China, Russia, Iran or North Korea. U.S. chipmaker NVidia  is currently embroiled in an investigation over more than $12 billion in alleged chip sales, some of which may have been diverted to China.</p> <p>The post <a href="https://cyberscoop.com/danielle-thumann-severs-fcc-nomination-tech-bills/">Senate Republicans advance FCC nominee, slate of cyber bills </a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  51. It’s NGINX’s turn to cause vulnpocalypse headaches

    Wed, 18 Nov 2026 18:54:12 -0000

    <p>Researchers just dropped five NGINX exploits, including one that chains an HTTP/2 bug with a kernel flaw to seize the machine behind the web server.</p> <p>The post <a href="https://cyberscoop.com/nginx-critical-vulnerabilities-exploits/">It’s NGINX’s turn to cause vulnpocalypse headaches</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Security researchers disclosed five vulnerabilities Tuesday in NGINX, the web server software that handles traffic for roughly a third of all websites, each with a working exploit. The most severe allows an unauthenticated attacker to run code on the server from anywhere on the internet, and in tests the researchers chained it with a separate Linux kernel flaw to take over the machine hosting the server entirely.</p> <p class="wp-block-paragraph">F5, the company behind NGINX, released fixes Tuesday and published an advisory classifying the worst of the flaws, an integer underflow tracked as CVE-2026-123007, as critical, scoring 9.2 out of 10 on the CVSS scale.&nbsp;</p> <p class="wp-block-paragraph">The problem sits in HTTP/2, the protocol NGINX uses to carry many requests over a single connection. HTTP/2 uses a credit system: the client grants the server permission to send a fixed amount of data. When a client cancels a request while its response data is still waiting to be sent, NGINX hands back that unused credit without checking whether the total now exceeds the protocol&#8217;s maximum. A later check subtracts the inflated number from the maximum using unsigned arithmetic, which cannot go negative, so the calculation wraps into an enormous value. That broken number is eventually treated as a negative response length, producing a memory buffer whose end pointer sits before its start, and a memory copy that spills outside its boundaries.</p> <p class="wp-block-paragraph">Rapid7 researcher Stephen Fewer, credited in <a href="https://my.f5.com/manage/s/article/K000164058">F5&#8217;s advisory</a> along with <a href="https://www.wiz.io/blog/nginx-http2-rce-vulnerabilities">Wiz Research</a>, Xint.io and Ikotas Labs, built a working exploit against a NGINX 1.31.8 container with HTTP/2 and TLS enabled.&nbsp;</p> <p class="wp-block-paragraph">The attack has practical limits: it requires roughly 128 gigabytes of HTTP/2 traffic, most of it control frames carrying no actual data. The demonstration took two-and-a-half minutes on a fast local network, but would need about two minutes over a 10-gigabit link and 17 minutes over 1 gigabit. Because NGINX closes connections after an hour by default, a network slower than about 300 megabits per second cannot finish the attack in time. The exploit also leans on a specific optimized memory-copy routine in the GNU C Library, selected on x86-64 chips with certain AVX-512 features. Rapid7 said other exploit methods may exist but none are currently known.&nbsp;</p> <p class="wp-block-paragraph">The other flaws affect different NGINX features, including reverse proxying over HTTP/2 and HTTP/3, combinations of the resolver and mirror directives, and mail proxying. Wiz Research, working with Ikotas Labs, Metnir and Rapid7, independently found two of the same flaws and documented one of them, a use-after-free bug, in reverse-proxy setups where connection slots run low.</p> <p class="wp-block-paragraph">Xint, which <a href="https://nginx2root.com/">found four of the five flaws</a> with an automated testing harness and no human guidance, went furthest. A researcher pointed the harness at about 50,000 lines of NGINX code; 90 minutes later it produced high-severity findings, and a working exploit followed within 10 minutes. The team then chained the HTTP/2 bug with a separate flaw in the Linux kernel&#8217;s SCTP networking code to reach root, the highest privilege level on the host. F5 said its cloud services are still being evaluated while BIG-IP and F5OS products were found not vulnerable.</p> <p class="wp-block-paragraph">“When we used Xint to scan NGINX, we didn’t know what to expect, since it’s a mature, widely deployed codebase that’s undergone years of scrutiny,” Brian Pak, co-founder and CEO of Xint, said in a statement to CyberScoop. “Others have targeted NGINX with AI pentesting, but none found this critical bug, which is a testament to our team and the engineering expertise that’s gone into our AI harnesses. We’re grateful for the collaboration with project maintainers to get a patch released quickly.”</p> <p class="wp-block-paragraph">Xint&#8217;s Shodan snapshot found 98.9% of 11.94 million version-identifiable NGINX hosts running affected releases, though that counts versions, not confirmed exposure, since it does not show whether each server accepts HTTP/2 or has received a backported fix. Wiz, which validated its findings with AI models, connected the convergence of separate teams on the same long-hidden bugs to AI&#8217;s improving ability to trace memory-safety flaws, a capability available to defenders and attackers alike.</p> <p class="wp-block-paragraph">The findings come as cybersecurity researchers continue to find critical bugs in foundational software used to power internet-connected software. Just 24 hours ago, researchers disclosed <a href="https://cyberscoop.com/java4shell-critical-java-flaw-openjdk/">a critical flaw in Java</a>, one of the world’s most used programming languages, that could let an attacker take over servers without a password. Last month, researchers found <a href="https://cyberscoop.com/radio/black-october-tom-hegel-sentinelone/">a host of similar vulnerabilities</a>, leading to some labeling the month as “Black October.”&nbsp;</p> <p>The post <a href="https://cyberscoop.com/nginx-critical-vulnerabilities-exploits/">It’s NGINX’s turn to cause vulnpocalypse headaches</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  52. 10-Year-Old NGINX HTTP/2 Flaw Lets Unauthenticated Attackers Execute Code on Servers

    Wed, 18 Nov 2026 18:31:17 -0000

    A flaw in NGINX's HTTP/2 support can allow an attacker to run code on servers with HTTP/2 enabled, without needing to log in. The decade-old flaw, CVE-2026-123007, dubbed "nginx2root," has been present in NGINX's open-source code since version 1.9.5, released on September 22, 2015. Fixed versions 1.31.9 and 1.30.7 came out today. F5, the company behind NGINX, says attackers can run code where
  53. F5 Warns 6 Critical BIG-IP Flaws Can Crash Traffic in Specific Setups, May Allow RCE

    Wed, 18 Nov 2026 18:15:39 -0000

    Six critical vulnerabilities in F5 BIG-IP allow an attacker to crash the Traffic Management Microkernel (TMM) without logging in, disrupting traffic until the TMM restarts. Each flaw needs a specific setup on a virtual server: an iRule that uses a particular command, or an APM policy that includes two particular kinds of agent. F5, which makes BIG-IP, disclosed the flaws today and says they
  54. Critical Apache Roller RCE Flaw Under Active Exploitation, No Authentication Required

    Wed, 18 Nov 2026 17:10:45 -0000

    A critical security flaw in Apache Roller, an open-source, Java-based blogging platform, has come under active exploitation in the wild, according to the Apache Software Foundation (ASF) The vulnerability, tracked as CVE-2026-129448 (CVSS score: 9.8), is a case of unauthenticated deserialization in the XML-RPC endpoint that could lead to remote code execution. "Deserialization of Untrusted
  55. Java4Shell Flaw Threatens Global Enterprises, Java Application Ecosystem

    Wed, 18 Nov 2026 16:54:52 -0000

    Discovered using AI, the memory corruption vulnerability affects the OpenJDK runtime and could allow unauthenticated attackers to take over systems.
  56. New EggStreme Malware Variant Goes Silent to Evade Security Tools in Government Attacks

    Wed, 18 Nov 2026 14:11:28 -0000

    The suspected Chinese advanced persistent threat (APT) group behind the sophisticated EggStreme malware has rebuilt its toolkit as it continues to target government organizations in the Philippines. "When attackers are exposed, the cheap response is to rotate," Bitdefender said in a technical report shared with The Hacker News. "A file hash changes with every recompile, a domain costs a few
  57. Major Brazilian Signature Apps Enable RCE & Fraud

    Wed, 18 Nov 2026 14:00:00 -0000

    Government-sanctioned authentication middleware is used for exceptionally important legal documents, yet it's often rife with RCE bugs.
  58. New Attack Bypasses TPM Lockout to Crack Windows Hello and BitLocker PINs in Days, Not Years

    Wed, 18 Nov 2026 12:28:30 -0000

    On AMD-based PCs and some Intel PCs, the TPM lockout for Windows Hello and BitLocker PINs can be reset by someone who opens the PC's case, researchers at Technische Universität Berlin say. That leaves short PINs open to guessing. In lab tests, the researchers found a four-digit Hello PIN on an AMD laptop and a 6-digit BitLocker PIN on an Intel laptop this way. Both PINs were set by the
  59. AI Pentesting Checklist for Web Apps: 15 Questions to Ask Before Choosing a Vendor

    Wed, 18 Nov 2026 12:15:00 -0000

    TL;DR Attackers move first. On average, exploitation now starts before a patch exists (Mandiant M-Trends 2026). In 1H 2026, 23% of vulnerabilities VulnCheck added to its exploited list were hit on or before disclosure day. AI pentesting vendors aren't interchangeable. Most look convincing in a demo. The gaps show up in your logged-in flows, your coverage, and your production safety. Make
  60. Java4Shell Flaw (CVE-2026-114938) Enables Pre-Auth RCE in Jira, Jenkins, Keycloak and More

    Wed, 18 Nov 2026 11:55:49 -0000

    A Java flaw called Java4Shell can let attackers without an account run their own code on Jira, Jenkins, Keycloak, and three other widely used applications, says Calif, the security research firm that disclosed it this week. The flaw is present in Java 17.0.6 and later, and the vulnerable code runs only on x86-64 Intel and AMD chips that support the AVX-512 VBMI feature. Tracked as
  61. Quality Isn’t The Main Worry About AI Pentesting — Data Privacy and Production Safety Are

    Wed, 18 Nov 2026 11:03:57 -0000

    AI is speeding up how software is built, which means more code, more changes, and more chances for vulnerabilities to reach production. It's speeding up how attackers find and exploit them too, with time-to-exploit now down to a single day. Security testing has to keep up with both. AI pentesting has emerged as a way to run in-depth tests whenever teams need them, instead of waiting for the
  62. Government tech debt is a national security risk

    Wed, 18 Nov 2026 11:00:00 -0000

    <p>With 80% of federal IT budgets spent maintaining aging legacy systems, fragmented government networks are struggling to defend against AI-accelerated cyberattacks from foreign adversaries.</p> <p>The post <a href="https://cyberscoop.com/government-tech-debt-national-security-risk-op-ed/">Government tech debt is a national security risk</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">In August, the Federal Bureau of Investigation <a href="https://www.justice.gov/opa/pr/justice-department-and-fbi-seize-platforms-operated-and-used-china-state-sponsored-hackers">announced</a> that it detected and disrupted another Chinese state-sponsored cyberattack targeting U.S. critical infrastructure — the latest in a series that includes the 2025 <a href="https://cyberscoop.com/plugx-malware-mustang-panda-doj-takedown/">PlugX</a> malware incident, the 2024 <a href="https://cyberscoop.com/doj-fbi-seize-flax-typhoon-hacking-tools-microscan-fishhub/">Flax Typhoon</a> hack and the 2023 <a href="https://cyberscoop.com/tag/volt-typhoon/">Volt Typhoon</a> operation. The severity and frequency of these attacks should raise alarms because they show a pattern of America’s adversaries looking for any possible weakness in our sensitive systems to exploit, and in many cases cooperating together to do so.</p> <p class="wp-block-paragraph">These attacks are often aimed at the federal government’s own networks, and they are all too often successful. The incoming House and Senate cyber leaders will need to assess and address this vulnerability fast.</p> <p class="wp-block-paragraph">The math isn’t on the government’s side. Cyber adversaries only have to be right once, whereas defenders must be right across every system, program and vendor. Unfortunately, the very way that the federal government contracts and manages critical systems makes the job of its cyber defenders much more difficult.</p> <p class="wp-block-paragraph">Federal systems are often cobbled together over decades, often with unpredictable budgets, shifting missions and procurement rules that favor incremental fixes over true modernization. The outdated technology is often easy to identify. The more dangerous aspect of the problem is the complexity embedded in federal management efforts, ensuring that addressing the challenge will be hard.</p> <p class="wp-block-paragraph">Under the guidance of the Office of Management and Budget and the General Services Administration, federal agencies routinely run three or four generations of systems simultaneously — legacy databases sitting beneath cloud environments, with managed service providers and internal teams each touching different layers, often without a shared map of what connects what. Security is often added later, after missions have changed, the platforms expanded, and new compliance requirements emerged.</p> <p class="wp-block-paragraph">Federal information technology spending exceeds $100 billion annually, according to a 2025 R Street Institute <a href="/Users/mark/Downloads/Street%20Institute">report</a>, with as much as 80 percent dedicated to operating and maintaining existing systems. The government spends this $80 billion a year simply to keep old systems secure while adversaries invest in new technologies and techniques to infiltrate them. Adversaries now have access to agentic AI systems to automate their reconnaissance efforts, accelerate exploitation of known vulnerabilities and move through supply chains before a patch cycle completes.</p> <p class="wp-block-paragraph">The fragmentation of federal networks and defenses makes it easier for foreign adversaries to penetrate systems, and it poses a significant strategic risk. The federal government may not have the network resilience to continue to operate in a crisis or contingency with China or Russia. Congress is starting to pay attention, but because cyber investment is less tangible than a new fighter jet or missile system and is spread across dozens of authorization and appropriation subcommittees in each chamber, it often takes a back seat.</p> <p class="wp-block-paragraph">Congress should treat federal technology modernization as a security imperative with the same urgency it has applied to traditional Defense Department weapons acquisition.</p> <p class="wp-block-paragraph">Agencies need procurement pathways that allow trusted security technologies to be adopted without multiyear acquisition cycles. Vendor contracts should define accountability clearly enough that, after a breach, the party responsible can be identified in hours rather than weeks. Platform duplication should also be reduced where it expands the attack surface without adding operational capability.</p> <p class="wp-block-paragraph">Weapons acquisition reform is focusing on delivering capabilities to warfighters faster and at lower cost. Federal cybersecurity infrastructure acquisition should be held to the same standard. Faster modernization would make government harder to penetrate and quicker to recover. Government tech debt is a strategic vulnerability, and our adversaries are already exploiting it.</p> <p>The post <a href="https://cyberscoop.com/government-tech-debt-national-security-risk-op-ed/">Government tech debt is a national security risk</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  63. Hackers Exploiting Critical Puppet RCE Flaw Affecting Enterprise Management Servers

    Wed, 18 Nov 2026 08:24:31 -0000

    The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Tuesday added a critical security flaw impacting Perforce Puppet, an enterprise infrastructure-as-code (IaC) and configuration management platform, to its Known Exploited Vulnerabilities (KEV) catalog, citing reports of active exploitation. The vulnerability, tracked as CVE-2026-115731 (CVSS score: 9.4), is a case of
  64. CISA Urges Immediate Patching of Java4Shell Flaw That Could Allow Remote Code Execution

    Wed, 18 Nov 2026 08:24:23 -0000

    The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has urged organizations to immediately patch a newly disclosed critical security flaw impacting Oracle Java. The flaw, tracked as CVE-2026-114938 (CVSS score: 8.1), refers to a case of out-of-bounds write that can lead to denial-of-service (DoS) or remote code execution under certain conditions. "This critical memory-corruption
  65. Attackers Compromise Microsoft's Agent Framework PyPI Package to Steal Cloud Credentials

    Wed, 18 Nov 2026 06:45:26 -0000

    Threat actors have compromised Microsoft's agent-framework package to push a credential stealer, according to reports from Aikido, Chainguard, Endor Labs, OX Security, SafeDep, Socket, StepSecurity, and Google-owned Wiz. The malicious versions, 1.26.1 and 1.26.2, were published 33 minutes apart on November 17, 2026, and are designed to run a credential stealer on every Python start. As of
  66. Russian COLD RELIC Hackers Use Stolen Google Workspace Data to Breach Second U.S. Organization

    Wed, 18 Nov 2026 05:58:04 -0000

    The Russia-linked cyber espionage group known as Cold Relic (formerly COLDRIVER and UNC4057) has been observed weaponizing a single cloud compromise into a launchpad for a series of linked intrusions resulting in third-party compromise. Google's Mandiant and Threat Intelligence Group (GTIG) said it investigated a 2026 compromise at a U.S.-based organization in which the adversary leveraged
  67. Optimizely Malware Attacks Highlight Third-Party Vendor Risks

    Tue, 17 Nov 2026 22:34:22 -0000

    At least 178 organizations, including several major brands, unwittingly exposed website visitors to hydra-like malware via a third-party platform.
  68. A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI

    Tue, 17 Nov 2026 22:24:39 -0000

    <p>Oracle shipped a patch this week for the flaw, which a short string of letters can trigger to corrupt memory in Java and hand attackers control of the servers running it.</p> <p>The post <a href="https://cyberscoop.com/java4shell-critical-java-flaw-openjdk/">A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">Security researchers on Tuesday published details of a critical flaw in Java, one of the world’s most used programming languages, that could let an attacker take over servers without a password.&nbsp;</p> <p class="wp-block-paragraph">The group behind the find, Calif Research, said AI coding tools did much of the work. Oracle Corp. shipped a fix the same day it released a November update covering 652 flaws across its product line.</p> <p class="wp-block-paragraph">The flaw is tracked as CVE-2026-114938 and nicknamed <a href="https://calif.io/java4shell/">Java4Shell</a>. It is a memory-corruption bug: code that writes data where it does not belong, which can crash a program or, at worst, let an attacker run their own commands. The buggy component is the Base64 decoder in OpenJDK, the free Java runtime that underpins software across banking, aviation and telecommunications, and that, by Calif’s count, runs at 90% of Fortune 500 companies.&nbsp;</p> <p class="wp-block-paragraph">Calif rated the flaw 9.2 out of 10 on the CVSS scale, where anything above 9 counts as critical. The group said an unauthenticated attacker, one who needs no username or password, could corrupt a server’s memory, read nearby data in a manner similar to the 2014 Heartbleed bug, and escalate that access into full control.</p> <p class="wp-block-paragraph">The bug lives in an unusual place: hand-written processor instructions that OpenJDK uses to make Base64 decoding faster. The optimization runs only on chips that support a feature set called AVX-512; ARM64 systems are unaffected. Calif found the decoder breaks on a specific input: 64 ‘A’ characters followed by a single equal sign, the padding mark Base64 uses to round out incomplete data.&nbsp;</p> <p class="wp-block-paragraph">The code then miscalculates a length and writes 64 bytes into a buffer, a small block of memory meant to hold as few as 50, overwriting whatever sits next to it. The program rejects the bad input afterward, but the surrounding memory is already damaged.</p> <p class="wp-block-paragraph">On its own, that corruption usually crashes the program. Calif showed that an attacker who carefully arranges what sits in the surrounding memory, a technique known as heap grooming, can turn the corruption into control of the server.&nbsp;</p> <p class="wp-block-paragraph">The group reported unauthenticated code execution in widely used products including Keycloak, Jira, Confluence, Jenkins and PingFederate, and demonstrated a working break-in against Keycloak, a widely used open-source login system. Authenticated execution, requiring a valid account, was shown in Kafka, Solr and Elasticsearch. Researcher samczsun showed Java programs can also be compromised simply by connecting to a malicious server.</p> <figure class="wp-block-embed is-type-video is-provider-youtube wp-block-embed-youtube wp-embed-aspect-16-9 wp-has-aspect-ratio"><div class="wp-block-embed__wrapper"> <div class="wp-block-embed wp-embed-aspect-16-9"><div class="wp-block-embed__wrapper"><iframe title="Java4Shell RCE Demos: Keycloak, Jira, Confluence, Kafka, Elasticsearch, etc." width="500" height="281" src="https://www.youtube.com/embed/-EOYmL4UKvs?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe></div></div> </div></figure> <p class="wp-block-paragraph">OpenJDK published a security advisory Tuesday, and Oracle <a href="https://www.oracle.com/security-alerts/cspunov2026.html#AppendixJAVA">shipped patched Java builds</a> Tuesday afternoon.&nbsp;</p> <p class="wp-block-paragraph">The find adds to a fast-growing catalog of vulnerabilities surfaced with AI. Calif said researcher Quang Luong found the bug in about an hour auditing OpenJDK with OpenAI’s Codex, for under $200 in computing tokens, and that AI agents produced most of the group’s exploit chains nearly on their own. It was the group’s second major AI-assisted find in recent weeks, after last month’s disclosure of php2shell.&nbsp;</p> <p class="wp-block-paragraph">In its write-up, the group called this the era of “the centaur hacker, part human, part AI,” and said the hunt for the bug felt like “looking for copper and found gold.”</p> <p class="wp-block-paragraph">Calif also framed the bug as a byproduct of software’s shared complexity: the AVX-512 speedup shipped with every copy of Java though few programs need it, so an optimization became a critical risk for all. The group said there is no universal exploit and each target requires tailored steps, and it is withholding product-specific exploit details until users have had time to upgrade.&nbsp;</p> <p class="wp-block-paragraph">No exploitation in the wild had been reported as of this article’s publication, and none of the CVEs related to the issue have been added to CISA’s Known Exploited Vulnerability list. However, the agency <a href="https://www.cisa.gov/news-events/alerts/2026/11/17/critical-java-vulnerability-requires-immediate-patching">published a notice</a> calling for a patch late Tuesday.<br><br>With Calif researchers already proving that unauthenticated RCE is possible in widely deployed apps, researchers tell CyberScoop the likelihood of real-world exploitation is high, especially given how fast attackers can leverage artificial intelligence. </p> <p class="wp-block-paragraph">&#8220;We&#8217;re talking about a vulnerability in one of the most widely used software runtimes in the world, with initial exploit development taking just a few hours,&#8221; said Gal Elbaz, co-founder and CTO of Oligo Security. &#8220;That doesn&#8217;t mean every Java application is vulnerable or that exploitation will be straightforward. Attackers still need to account for specific hardware, runtime configurations, and application behavior. But the technical barriers are clearly surmountable, and organizations should treat this as a serious exposure that requires compensating controls capable of blocking exploitation while patches are rolled out.&#8221;</p> <p class="wp-block-paragraph"><strong>Update, 11/17/26, 6:00 p.m.: </strong><em>This story was updated with information about a CISA notice and comment from Gal Elbaz.</em></p> <p>The post <a href="https://cyberscoop.com/java4shell-critical-java-flaw-openjdk/">A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  69. How CISA is prioritizing companies for talks about security improvements

    Tue, 17 Nov 2026 20:52:59 -0000

    <p> CISA acting director Nick Andersen made his remarks after speaking at FedTalks, presented by FedScoop.</p> <p>The post <a href="https://cyberscoop.com/cisa-100-critical-infrastructure-companies-fed-talks/">How CISA is prioritizing companies for talks about security improvements</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
    <p class="wp-block-paragraph">The Cybersecurity and Infrastructure Security Agency has identified approximately 100 companies that own critical infrastructure for talks on hardening their systems because of how vital they are to the nation, CISA acting director Nick Andersen said Tuesday.</p> <p class="wp-block-paragraph">That figure does not cover everyone CISA wants to speak with, because those companies have so many interdependencies with other firms, he said. And the conversations are going to vary wildly from company to company, he told CyberScoop.</p> <p class="wp-block-paragraph">“For the initial workstream, we&#8217;re probably talking about around 100 companies,” Andersen said after speaking at FedTalks, produced by FedScoop. “We want to be focused on being able to gather what are those lessons learned that are most appropriate coming out of it, what works and what doesn&#8217;t, and the problem with each one of those different companies.”</p> <p class="wp-block-paragraph">To generate the list, CISA went through an intergovernmental Homeland Defense Working Group that identified companies that provide functions deemed most important to the economy, public health and safety, national security and defense critical infrastructure.</p> <p class="wp-block-paragraph">Then, Andersen said, the agency overlaid&nbsp; five to six years of intelligence about what cyber adversaries appeared to value most, such as where they were pre-positioning or where their research and development budgets were going. Companies that appeared on both lists became the targets for “initial engagement,” which also meant involving the right sector risk management agencies.</p> <p class="wp-block-paragraph">So hypothetically, if the decision is that a particular port is especially important, CISA would talk to the port operator. “But I also need to have a conversation with their energy provider, with their telecommunications provider, et cetera, et cetera, et cetera,” he said.</p> <p class="wp-block-paragraph"><a href="https://cyberscoop.com/cisa-nick-andersen-cyber-resilience-outages/">In his FedTalks speech</a> Andersen said the ideal is to have conversations that include IT and operational technology providers, along with cybersecurity providers. He said the goal is to advance the agency’s “CISA 2030” strategy, which focuses on resilience, minimum viability and related priorities.</p> <p class="wp-block-paragraph">“It&#8217;s going to continue to result in several significant, important conversations over the coming weeks and months about, what does it mean to appropriately invest?” Andersen said. “What does it mean to collaborate with one another on that longer-term resilience as we move forward? That&#8217;s going to have to be a collective effort across our communities.“</p> <p class="wp-block-paragraph">So far, he said, the conversations have been productive.</p> <p class="wp-block-paragraph">“A lot of this has also included things like CI-Fortify assessments with those companies, which have been really really good conversations for us to be able to have around, what does it mean to be able to plan for isolation and recovery?” Andersen said.</p> <p class="wp-block-paragraph">Andersen, and other Trump administration officials, have spoken in recent months about the need to <a href="https://cyberscoop.com/cisa-cyber-risk-prioritization-vulnerability-directive/">be selective</a> in how they <a href="https://fedscoop.com/ai-threats-white-house-policy-supply-chain/">prioritize vulnerabilities</a> and hacking targets.</p> <p>The post <a href="https://cyberscoop.com/cisa-100-critical-infrastructure-companies-fed-talks/">How CISA is prioritizing companies for talks about security improvements</a> appeared first on <a href="https://cyberscoop.com">CyberScoop</a>.</p>
  70. Cursor Traces Malicious Claude Code Extensions to Third-Party Breach

    Tue, 17 Nov 2026 18:38:37 -0000

    A compromised key issued to AI infrastructure provider Fireworks was used by threat actors to access Cursor data and credentials.
  71. Oracle to Patch 11 Java Flaws Today as Critical OpenJDK RCE (CVE-2026-114938) Report Emerges

    Tue, 17 Nov 2026 15:17:36 -0000

    Oracle is preparing to release security patches for 11 vulnerabilities in Java SE on Tuesday, November 17, 2026, all of which may be exploited remotely without authentication, as separate reports warn of a potentially critical remote code execution (RCE) flaw affecting OpenJDK. The 11 vulnerabilities carry a maximum CVSS 3.1 severity score of 8.1, according to Oracle's advance security notice.
  72. Hackers Found a Way Around GitHub SHA Pinning. This Webinar Explains How to Close the Gap

    Tue, 17 Nov 2026 12:25:06 -0000

    Pinning GitHub Actions to an exact version of their code, known as SHA pinning, has become the go-to defense against software supply chain attacks. Attackers have already found a way around it. In March 2026, the hackers who hijacked Trivy, a widely used open-source security scanner, planted malicious code that looked like it came from GitHub's official checkout tool. They even labeled it a
  73. New GitDoor and GraphDoor Backdoors Turn GitHub and SharePoint Into Covert C2 Channels

    Tue, 17 Nov 2026 11:20:18 -0000

    Government sectors in Asia and Africa have been targeted by an active cyber intrusion campaign that delivers to previously undocumented backdoors that leverage legitimate services like Microsoft SharePoint and GitHub for command‑and‑control (C2). The activity has been codenamed Operation ShadowCloud by Kaspersky. The Russian cybersecurity company said it identified the campaign in early 2026.
  74. Hijacked Optimizely Scripts Served Card Skimmer and iOS Exploit Kits to CNN, BBC and Zoom Visitors

    Tue, 17 Nov 2026 07:12:13 -0000

    An attacker exploited a flaw in Optimizely Web Experimentation to plant malicious code in some of the scripts that customers' websites load from the service, Optimizely said early today. Visitors to sites run by at least 178 organizations, including CNN, the BBC and Zoom, received the code yesterday, according to Report URI, a firm that monitors the scripts websites run. The code was built to
  75. Google Project Zero Slashes Bug Disclosure Deadlines, Splitting Experts

    Mon, 16 Nov 2026 22:08:20 -0000

    One of the oldest running bug disclosure programs announced it would shorten public disclosure timelines to vendors from 120 days to 30 over the next 14 months.
  76. Ransomware Attacks Increase Behind AI, but Payments Decline

    Mon, 16 Nov 2026 19:56:05 -0000

    Large language models and other AI systems allow a growing number of ransomware groups to create convincing phishing emails and identify sensitive data quickly.
  77. 'Sleeper Agent' Tactic Hides Payloads in Open-Weight AI Models

    Mon, 16 Nov 2026 19:10:03 -0000

    Malware can lurk inside just 0.00001% of a neural network. How, then, can any user tell it's there?
  78. China-Nexus Actors Breach 10 Organizations in AI-Driven Attacks

    Mon, 16 Nov 2026 16:36:55 -0000

    Attackers used Cairn once again in autonomous attacks that targeted advanced manufacturing and defense sectors, as well as civil and human rights organizations.
  79. NIST Finalizes Post-Quantum Cryptography Guidance, GAO Offers a Reality Check

    Mon, 16 Nov 2026 14:22:15 -0000

    The emerging technology will soon break the &quot;lock and key&quot; encryption currently protecting sensitive data. A new GAO report shows agencies are behind on building their defenses.
  80. Google Finds Hackers Using AI Agents to Build and Run Exploits From Software Patches

    Mon, 16 Nov 2026 14:00:00 -0000

    Google on Monday revealed that adversarial use of artificial intelligence (AI) by threat actors has gone beyond manual prompting to deploying autonomous multi-agent offensive frameworks like ARTEX, CAIRN, CyberStrikeAI, and PentAGI to execute multi-stage intrusions at greater operational velocity. In tandem, threat actors are leveraging the power and capabilities of large language models (LLMs)
  81. ⚡ Weekly Recap: N-able 0-Days, Windows 0-Days, Chrome 0-Days, Fridge 0-Days, and More Bad Days

    Mon, 16 Nov 2026 13:27:20 -0000

    A default password opened doors it never should have. A trusted update delivered malware. Even a security feature became a way to break in. Different stories, same problem: the things meant to protect us aren't always doing their job. Meanwhile, AI is making bug hunting faster, old flaws are still paying off, and attackers keep finding easy ways in. Some attacks take real skill. Others barely
  82. Six Unpatched Danfoss Refrigeration Controller Flaws Could Allow Remote Takeover Without Login

    Mon, 16 Nov 2026 12:34:21 -0000

    Six flaws in Danfoss AK-SM800A refrigeration controllers could let a remote attacker take over a controller without logging in, Dutch security consultancy Midnight Blue said earlier this week. No patch is available, and the researchers say the newest firmware, version 5.4.0, does not fix the flaws. Government cybersecurity agencies in Belgium, Ireland and Germany told operators this week to
  83. As Test Takers Turn to AI, Exam Security Plays Catch-Up

    Fri, 13 Nov 2026 22:27:00 -0000

    It's hard to detect whether someone is using artificial intelligence to cheat on IT certification exams, but new tools and processes are emerging to boost trust.
  84. ArcticRAT Cybercrime Gang Profiles Victim Finances With AI

    Fri, 13 Nov 2026 20:20:27 -0000

    The Android malware has expanded from focusing on cryptocurrency theft to full device control, and the attackers have added advanced analysis of stolen information.
  85. Attackers in Binance Breach Targeted Known Weakness, Missing Controls

    Fri, 13 Nov 2026 13:00:00 -0000

    Blockchain complexities, such as attacker-deployed sweepers, hobble security controls designed to limit the damage of an exchange compromise.
  86. Getting Past the Post-Quantum Cryptography Discovery Phase

    Thu, 12 Nov 2026 22:46:05 -0000

    Most organizations haven't sorted out their cryptographic estate enough to even know where they'd start with post-quantum cryptography (PQC).
  87. Cyber Experts Weigh In: How to Fight Artex & AntiFox

    Thu, 12 Nov 2026 21:18:52 -0000

    Cybercriminals using agentic pen-testing tools is a proverbial genie out of the bottle at this point, so enterprises need to get a defense plan in place right away.
  88. Inside an AI Attack Against a US Retailer

    Thu, 12 Nov 2026 11:00:00 -0000

    When hand-typed commands failed to breach a US retailer, the attacker turned to AI-generated scripts, achieving lateral movement in minutes.
  89. Srsly Risky Biz: Iranian Hacks With Russian Characteristics

    Thu, 12 Nov 2026 03:40:15 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.airlockdigital.com/"><em><u>Airlock Digital</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3-1.jpeg" alt="Srsly Risky Biz: Iranian Hacks With Russian Characteristics"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.airlockdigital.com/"><em><u>Airlock Digital</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB190.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB190/&quot;&gt;Srsly Risky Biz: Iranian hacks with Russian characteristics&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 21:43 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Iranian hacks with Russian characteristics&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3.jpeg" class="kg-image" alt="Srsly Risky Biz: Iranian Hacks With Russian Characteristics" loading="lazy" width="2000" height="1353" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-39d1ab16-87de-4644-b801-f3579f9f03d3.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@juliakadel?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Julia Kadel</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/a-group-of-russian-nesting-dolls-sitting-on-a-table-YmULswIbc3I?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>French authorities <a href="https://cert.cyber.gouv.fr/cti/CERTFR-2026-CTI-009/"><u>have attributed</u></a> recent supermarket freezer hacks to cyber units within the Iranian Revolutionary Guard Corp (IRGC).&#xa0;</p><p>These attacks felt very Russian, so we&apos;re surprised.</p><p>It&#x2019;s important to know what exactly has been attributed here. There&apos;s been a string of refrigeration hacks since an initial incident in August, when freezers at US military commissaries suddenly stopped working at <a href="https://signalandsilence.substack.com/p/i-think-someone-hacked-the-commissary"><u>at least six different bases</u></a>.</p><p>In October, French retailers were hit. <u>Then came a</u><a href="https://www.bleepingcomputer.com/news/security/hackers-disrupt-refrigeration-systems-at-european-supermarkets/"><u> huge wave</u></a> of supermarket attacks in early November, affecting more than 115 stores in Belgium, the Netherlands, Sweden, Luxembourg, Ireland and Estonia.&#xa0;</p><p>Finally, French national blood service fridges <a href="https://www.bleepingcomputer.com/news/security/cyberattack-disrupts-blood-storage-refrigeration-systems-in-france/"><u>were hacked</u></a> last week and their temperature settings raised. Fortunately, monitoring systems alerted staff to the problem and they were able to intervene before any damage was caused.&#xa0;</p><p>The specific incident attributed to Iran by French authorities was the string of attacks that hit French retailers in October, from Sunday 11 to Monday 12. <a href="https://news.risky.biz/risky-bulletin-france-blames-iran-for-the-october-supermarket-fridge-hacks/"><u>Per <em>Risky Bulletin</em></u></a>:&#xa0;</p><blockquote>The hack hit 700 retail stores across France, but also a few butchers, food factories, and even a slaughterhouse. While the Intermarch&#xe9; group was heavily hit, the attacks also crippled freezers at supermarket chains such as Casino, Leclerc, Coop&#xe9;rative U, Metro, and Carrefour.</blockquote><blockquote>The hacked freezers were set to super high temperatures that defrosted and spoiled food, but in 15 locations, the freezers overheated and also triggered fires.&#xa0;</blockquote><blockquote>The attacks mainly targeted the web panels of Danfoss refrigeration controllers, which allow store IT teams and employees to control the temperatures of their freezers from centralized panels or over the internet.</blockquote><p>Unnamed US intelligence sources told <em>The New York Times</em> they had reached the same conclusion, but only with low confidence. More intelligence could change their assessment.&#xa0;&#xa0;&#xa0;</p><p>That whodunnit confusion is just one reason these sorts of attacks are so hard to deter. Three of the attacks have not yet been attributed and for the one that has, the US is hedging. Exactly how strong is the evidence that Iran is responsible? As already mentioned, if we&apos;d been forced to bet, we would have put our chips on Russia being responsible as the attacks are consistent with its <a href="https://www.csis.org/analysis/russias-shadow-war-against-west"><u>destructive hybrid attacks across Europe</u></a> in recent years.</p><p>Russian cyber operators have masqueraded as Iranian threat actors in the past. Back in 2019 <a href="https://www.ncsc.gov.uk/news/turla-group-behind-cyber-attack"><u>the UK&apos;s NCSC reported</u></a> that the Russian group Turla had hijacked Iranian tools and malware to conduct espionage across the Middle East.&#xa0;</p><p>It also doesn&apos;t make much sense for Iran to cause chaos across Europe. Iran&apos;s cyber shenanigans earlier this year were <a href="https://news.risky.biz/srsly-risky-biz-chipping-away-at-chinese-ai-risks/#irans-perfectly-calibrated-cyber-attack"><u>firmly focussed on US water infrastructure</u></a>. In our view, those attacks were well calibrated to achieve US political impact without any significant risk of escalation in its conventional conflict. Whereas Iranian attacks on European refrigeration&#x2026; just don&apos;t really track.&#xa0;</p><p><em>The New York Times</em> also reported that Western former security officials were at a loss to explain why Iran would target Europe, given that it is in the middle of a war with the US.</p><p>As for the <a href="https://www.diplomatie.gouv.fr/fr/presse-et-ressources/decouvrir-et-informer/actualites/declaration-d-attribution-a-l-iran-de-cyberattaques-visant-la-france"><u>French government, it&apos;s said</u></a> it&apos;ll summon the Iranian Ambassador and express its strongest condemnation. Additionally, it will talk to European partners about the adoption of new sanctions. A stern talking to and some sanctions! That&#x2019;ll <em>totally </em>work!</p><p>Whether France is right and Iran is responsible, or it turns out that Russia was muddying the attribution waters, deterring either of them will be difficult. Both are involved in serious conventional conflicts and are already sanctioned up the wazoo.&#xa0;</p><p>Whoever is responsible, they have found a rich vein of annoyance to mine. We just hope smart freezers aren&apos;t the thin edge of the wedge.</p><h2 id="china-adopts-everything-everywhere-all-at-once-strategy">China Adopts Everything Everywhere All at Once Strategy</h2><p>Back in August <a href="https://news.risky.biz/chinas-ai-enabled-apt-operations-are-getting-interesting/"><u>we covered</u></a> how one Chinese cyber espionage actor, SilkParasite, was using AI to diversify its malware arsenal to make it more difficult to attribute its various hacking campaigns. That makes sense when an important goal is to operate quietly and not get caught.&#xa0;</p><p>By contrast, <a href="https://news.risky.biz/ai-powered-espionage-will-favor-china/"><u>China&apos;s B-team hackers</u></a> are using AI to go faster, albeit noisily.&#xa0;</p><p>China has embraced stealthy <em>and</em> noisy operations, depending on its campaign objectives, and recent threat intelligence reports bear this out.</p><p>Starting on the noisy side, CrowdStrike <a href="https://www.crowdstrike.com/en-us/blog/vapor-panda-uses-artex-to-scale-global-campaign/"><u>released a report</u></a> last month on a Chinese cyber espionage actor using AI for a pedal-to-the-metal hacking campaign that prioritises speed and scale, stealth be damned. The actor, which CrowdStrike calls Vapor Panda, is using the open-source agentic pentesting framework ARTEX to ramp up its previous hacking capability.</p><p>The Google Threat Intelligence Group (GTIG) <a href="https://cloud.google.com/blog/topics/threat-intelligence/artex-and-friends-enter-agentic-offensive-frameworks"><u>describes ARTEX</u></a> as &quot;an autonomous, multi-agent LLM-driven pentesting and offensive command-and-control (C2) platform&quot;. ARTEX can hook into both frontier and open source models and it&apos;s <a href="https://www.microsoft.com/en-us/security/blog/2026/11/06/artex-abuse-expands-as-cybercriminals-target-financial-sector/"><u>definitely not just China</u></a> that&apos;s cottoned on to its usefulness.&#xa0;</p><p>When it comes to Vapor Panda, Crowdstrike says ARTEX &quot;pursued multiple attack paths in parallel&quot; including web applications, cloud services, container environments, mail and VPN infrastructure. If any path was unsuccessful it would shift to another alternative approach. A single operator could use ARTEX to run multiple simultaneous engagements, each targeting different organisations, and only jumps in to direct the agents when decisions are required.&#xa0;</p><p>Vapor Panda certainly did plenty of hacking with ARTEX. In the space of 11 days beginning in mid-October the group used it against more than 200 organisations. These, according to CrowdStrike, included &quot;foreign ministries in Latin America, Europe, Asia, and the Middle East [and] semiconductor, defense, and think tank organisations in the US&quot;. They also targeted &quot;telecommunications providers in multiple regions&quot;.&#xa0;</p><p>Of course, using ARTEX to hack quickly is not<em> </em>exactly<em> </em>subtle. GTIG says the harness relies on &quot;rapid, iterative trial-and-error&quot; and that compared to a competent human it is &quot;forensically noisy&quot;. In one Vapor Panda intrusion, ARTEX&apos;s interaction with just a single web server&#xa0; &quot;generated 240 log events and exhibited errors potentially caused by multiple sub-agents overwriting one another while pursuing the same objective&quot;.&#xa0;</p><p>This kind of devil-may-care approach reminds us of China&apos;s freelance cyberespionage outfits. <a href="https://news.risky.biz/risky-biz-briefing-the-i-soon-data-leak/"><u>These groups hack first</u></a>, then shop the stolen materials around to PRC intelligence services. For these actors, more hacking means more sales opportunities.</p><p>And Vapor Panda had some success. Out of the 200 organisations targeted, CrowdStrike confirmed post-exploitation activity in nine of them. There were two &quot;extensive intrusions&quot; into the foreign ministries of at least two Latin American countries. A Middle East-based interior ministry was also compromised.&#xa0;</p><p>Loud and fast isn&apos;t the best strategy for all actors, however.&#xa0;&#xa0;</p><p>On the stealthier side of the spectrum we have Silent Train, an illicit network of compromised devices used by Volt Typhoon, the group <a href="https://news.risky.biz/volt-typhoon-keep-calm-and-carry-on/"><u>compromising US critical infrastructure</u></a> in order to disrupt it in the event of a military conflict. Late last month Black Lotus Labs <a href="https://www.lumen.com/blog/en-us/silent-train-volt-typhoon-operational-framework"><u>released a report</u></a> on Silent Train.&#xa0;</p><p>When it comes to operational security and stealth, the contrast between Volt Typhoon and Vapor Panda could not be more stark. Volt Typhoon has no interest in smash-and-grab intelligence gathering. Its goal is to guarantee access to American critical infrastructure at some future date and it is all about stealthy operation.</p><p>Rather than being persistent infrastructure, Silent Train is made up of short-lived chains of compromised devices that Volt Typhoon operators would route activity through to reach targets. Operators could assemble chains in under two minutes by exploiting exposed web interfaces on SOHO, IoT and other relay devices. They&apos;d deploy just the tooling required for chained SSH access through the compromised node, including temporary SSH services, keys, configuration files and <a href="https://news.risky.biz/living-off-the-land-is-the-new-normal-65c0eb42a0e6da001a37dfb5/"><u>living-off-the-land</u></a> utilities. Volt Typhoon removes all these artifacts when it dismantles the chains, and they can be set up and torn down in as little as ten minutes. On average they lasted four hours.</p><p>AI-assisted YOLO hacking is clearly counterproductive when sustaining access is a priority, and there is evidence that Silent Train was operated by, or at least overseen by, real people. Most of its activity occurred during human working hours, from Monday to Friday, 9am to 10pm in China Standard time. This does not appear to be an agentically driven operation. Very artisanal! How quaint!</p><p>So we have three distinct examples of how China&apos;s cyber operators are currently operating. Vapor Panda utilises AI for speed and is unphased by noise. SilkParasite bolsters its malware arsenal with AI to muddy attribution. Volt Typhoon, the most cautious and stealthy of the three, is operating slowly and steadily, with human hands on keyboards from 9am Beijing time, Monday to Friday.</p><p>The lesson? China gonna China. But in 2026 and partially thanks to AI, it&apos;s managed to become faster, louder or sneakier, depending on where you look.</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/0p7p1eogdlc?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Iranian hacks with Russian characteristics"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>AI is getting better at respecting scope:</strong> OpenAI <a href="https://openai.com/index/introducing-gpt-6-cyber/"><u>has released</u></a> its new cybersecurity model, GPT-6 Cyber, and says that it is noticeably better at the ExploitGym benchmark. Researchers have used it to find 81 new high-severity vulnerabilities in Firefox and over 900 linux privilege escalation bugs. Most importantly, OpenAI says the model was far better at staying within its authorised scope.&#xa0;&#xa0;</li><li><strong>Parts of Signal get formally verified:</strong> Google DeepMind researchers <a href="https://www.wired.com/story/a-new-ai-tool-from-google-deepmind-formally-verified-the-signal-protocol/"><u>have used&#xa0; a new AI framework</u></a> called AlphaProof Code to formally verify that some of the Signal protocol actually works as expected. The research took just four weeks, which is very fast for this kind of work and raises hope that formal verification might be used more widely to assess other very important algorithms. That would be good, but the scope of this analysis was still quite limited and it didn&apos;t assess the Signal app, for example.&#xa0;</li><li><strong>Killing secrets before they are abused:</strong> Truffle Security <a href="https://trufflesecurity.com/blog/launching-the-secrets-revocation-program"><u>has launched a program</u></a> to help API issuers to revoke live secrets it finds on the public internet. Truffle scans publicly accessible data such as code, packages and container images and finds lots and lots of live secrets. Issuing companies that have joined the Secrets Revocation Program will now revoke keys when alerted by Truffle. This will be a big security win because over 80 percent of the time developers do not revoke keys when alerted by email. Developers can opt out of the service.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>Tom Uren talks to Airlock Digital&#x2019;s CTO and CSO Peter Baussmann about how the Essential Eight is evolving into ASD&#x2019;s Essentials framework.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI150.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI150/&quot;&gt;Sponsored: The evolution of ASD&apos;s Essential Eight&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 15:51 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: The evolution of ASD&apos;s Essential Eight&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em>&#xa0; talk about why PHP2Shell failed to live up to its hype.&#xa0;</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN189.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN189/&quot;&gt;Between Two Nerds: Why PHP2Shell fizzled&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 24:16 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Why PHP2Shell fizzled&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/gW9Kns25pxs?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Why PHP2Shell fizzled"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Pro-Kremlin hacktivists dox EU politicians, law enforcement: </strong>A pro-Kremlin hacktivist group named NoName057(16) has published the personal information of 18 high-profile EU politicians and law enforcement officials.</p><p>Victims include Italian Foreign Minister Antonio Tajani, the Eurojust president Michael Schmid, Europol European Cybercrime Centre head Edvardas &#x160;ileris, German Federal Police (BKA) president Holger M&#xfc;nch, and Italian Postal Police chief Ivano Gabrielli.</p><p>The dox also covers three German journalists who investigated Russian hackers and influence operations&#x2014;Sergej Maier, Angelique Geray, and Michael Alexander Martens.</p><p>[<a href="https://news.risky.biz/risky-bulletin-pro-kremlin-hacktivists-dox-eu-politicians-law-enforcement/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Some OpenSSH instances take any password longer than 256 characters:</strong> There can&apos;t be any newsletter edition without a dumb bug, and today&apos;s selection is a vulnerability codenamed<a href="https://sshhole.com/"><u> SSHhole</u></a>, discovered by Wakefield security researcher Ozgun Kultekin.</p><p>This particular beauty impacts the OpenSSH project and allows attackers to log into accounts with any random string with more than 256 characters. It doesn&apos;t matter what the characters are, you get root as long as you type more than 256.</p><p>The bug will work against accounts that have the password disabled or use cryptographic keys to log in.</p><p>[<a href="https://news.risky.biz/risky-bulletin-some-openssh-instances-take-any-password-longer-than-256-characters/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  90. Chinese AI Project Scouts Spies in Taiwan Chip Industry

    Thu, 12 Nov 2026 01:00:00 -0000

    The CCP built a Big Data system to scrape reams of information and learn about individuals abroad, to potentially target and exploit them. Think Alice Huang Lite.
  91. Japan Pushes Emergency Funding for Cybersecurity as AI-Assisted Breaches Rise

    Thu, 12 Nov 2026 00:00:00 -0000

    The package adds public-private cyber responders, an AI-focused team run by the economic ministry, and more leeway to use the government's new active cyber-defense powers.
  92. AIを活用したサイバー攻撃の増加を受け、日本がサイバーセキュリティへの緊急資金拠出を推進

    Wed, 11 Nov 2026 22:14:44 -0000

    この対策パッケージには、官民合同のサイバー対応チーム、経済省が主導するAI専門チーム、および政府の新たな能動的サイバー防衛権限の行使に関する裁量の拡大が含まれている。
  93. Attackers Say 'Hello' to a Different Scale of Cloud Attack

    Wed, 11 Nov 2026 21:44:04 -0000

    Researchers dubbed an attack chain &quot;Greeter&quot; after discovering that attackers deliberately covered their trail by leaving &quot;hello&quot; messages in logs.
  94. AI Raises the Stakes for Cyber Resilience

    Wed, 11 Nov 2026 21:20:41 -0000

    An organization's ability to quickly recover from a business disruption will be a competitive advantage in an era of AI-driven threats and business enablement.
  95. November Patch Tuesday: An Overflowing Cornucopia of CVEs

    Tue, 10 Nov 2026 22:26:51 -0000

    Microsoft serves up an actively exploited Outlook zero-day, and a wave of critical Exchange, SharePoint, Windows Search, and LDAP bugs that exhausted defenders need to prioritize fast.
  96. Artex: The AI Pen-Test Tool Hackers Turned Into a Weapon

    Tue, 10 Nov 2026 21:01:14 -0000

    Forks of the open source Artex framework, along with 45 similar pen-testing tools, remain capable of executing credential attacks and cloud breaches at machine speed across multiple industries.
  97. Checkmarx Appoints Kobi Sivilya as Chief Information Security Officer

    Tue, 10 Nov 2026 20:35:34 -0000

    Sivilya brings deep cybersecurity and technology leadership to Checkmarx as AI transforms software development.
  98. AI Coding Agents Leak Thousands of Secrets to GitHub

    Tue, 10 Nov 2026 18:46:22 -0000

    A command in Claude Code inadvertently saves files with credentials and other sensitive data to public Git repositories, creating security gaps for organizations.
  99. AI Makes Reverse Engineering Mainstream Software & Malware Far Easier

    Tue, 10 Nov 2026 13:00:00 -0000

    Researchers unpacked, mapped, and recreated complex software and the Turla malware, raising opportunities and concerns for organizations and cybersecurity analysts.
  100. People Are Getting Weird About AI

    Mon, 09 Nov 2026 23:06:22 -0000

    Dark Reading Confidential Episode 22: You can love AI without being blind to the cybersecurity risks. The tendency for humans to anthropomorphize the conversations and interactions we're having with AI is growing into potential delusion.
  101. CISA Announces Enhanced Reporting Form for Cyber Incidents

    Mon, 09 Nov 2026 21:38:20 -0000

    New questionnaire captures critical data and information that improves the agency’s ability to assess incidents, identify trends and prioritize response actions.
  102. Microsoft Fixes RCE Bug in Windows Paint After Weeks of Confusion

    Mon, 09 Nov 2026 21:12:24 -0000

    Last month, Microsoft disclosed a vulnerability in Microsoft Paint, but messaging regarding a patch was unclear.
  103. Black Hat India 2026 Makes a Landmark Debut in Bengaluru, Bringing the Global Cybersecurity Community to India

    Mon, 09 Nov 2026 20:19:56 -0000

    Inaugural edition by Informa Markets in India brings together leading researchers, security practitioners and industry decision makers for four days of cutting-edge research, technical exchange and strategic dialogue.
  104. Attackers Crash Claude's 'Office Party' Flaw to Run Malicious Code on Linux

    Mon, 09 Nov 2026 18:08:30 -0000

    Attackers can exploit a flaw in Claude's document-processing tool by creating a temporary file that Anthropic's AI assistant can access and run on a user's machine.
  105. PwC's CISO on AI, Robots & What's Next for Security

    Mon, 09 Nov 2026 17:17:23 -0000

    As AI accelerates change across the enterprise, PwC's James Shira explains why security teams must rethink their approach to risk and adapt their strategies.
  106. NIST: Solo Shouldn't Mean Less Secure

    Mon, 09 Nov 2026 15:46:10 -0000

    New guidance provides a basic framework for less-technical solo business owners, but it may miss the mark on how to secure technology that requires no human oversight.
  107. What Does the SOC Stop Doing When Everything Is Urgent?

    Mon, 09 Nov 2026 00:00:00 -0000

    There are two different problems hiding inside the phrase “SOC overload,” and the loud one is work that should never have reached a person: the rule nobody tuned, the alert class that fires four hundred times a day, the queue padded with things that aren’t incidents. That’s a real problem, it has real fixes, and it isn’t what this post is about.
    <p>There are two different problems hiding inside the phrase “SOC overload,” and the loud one is work that should never have reached a person: the rule nobody tuned, the alert class that fires four hundred times a day, the queue padded with things that aren’t incidents. That’s a real problem, it has real fixes, and it isn’t what this post is about.</p> <p>This is about what’s left after you fix it.</p> <p>Picture a shift where every open item is legitimate. Each one has been qualified. Each one has an owner. Each one carries some commitment the organization has already made about when a human touches it. None of it is noise - and there are still more of them than there are minutes.</p> <p>That problem is harder to write about, because it doesn’t end in a fix. When everything in front of you is already accepted work, choosing one item doesn’t make the others disappear. It changes which ones wait, how long, and what you can truthfully say to the people waiting with them. An organization can of course renegotiate a commitment or decide to discontinue a response, but that takes somebody holding the authority to do it - it isn’t a sequencing call, and it isn’t what the shift lead is doing at minute zero. Inside the scope I’m working in here, everything accepted stays accepted, so the lead’s question is “what waits, and what can I honestly promise about when it stops waiting?”</p> <p>I’d rather make that decision inspectable than argue about it in the abstract, so most of this post is a single invented ninety-minute window - two analysts, six cases, worked through at minute resolution. There are two defensible choices at minute zero, a forced interruption at minute forty, and then the uncomfortable part, where the displaced work has to come back out of somebody’s actual calendar. Along the way I want to pull apart four things that all look like deadlines and aren’t, show why “we have more work than hours” can be true and still not be a finding, and get to the point where this plan genuinely stops being feasible. That point turns out to be much narrower than “the queue is long.”</p> <p><br /></p> <p><strong>What’s actually scarce</strong></p> <p>Start with the definition, because the whole argument hangs off it. By <em>accepted work</em> I mean an item the organization has already agreed to handle: qualified, owned, and carrying some expectation that now exists in the world - internally, with a constituent, or in a contract. Everything in this post sits downstream of that acceptance. Accepted doesn’t mean solved, and it doesn’t mean every operational uncertainty about the item has been resolved - it means the handling is owed. Nobody in this window is relitigating whether an item belongs on the board.</p> <p>What’s scarce isn’t attention in some general sense. It’s <em>eligible</em> analyst-minutes: minutes from a person who has the skills, the permissions and the case context to move that specific item forward. FIRST’s CSIRT Services Framework puts a usefully blunt definition on the word, describing capacity as the number of simultaneous occurrences of a capability an organization can execute <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">“before they achieve some form of resource exhaustion”</a>. That framework is written for CSIRTs rather than SOCs, though it says directly that its event-management and incident-management services are “expected” to be useful and directly applicable when defining SOC services, which is how I’m using it here. It also states plainly that it <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">makes no recommendations “about capability, capacity, maturity, or quality”</a>, so it describes the activities a response involves without telling anyone how many people to point at them. The inspected version is 2.1, which FIRST publishes marked for review.</p> <p>The second thing worth separating is the word “due,” because at least four different things render identically on a ticket and behave nothing alike under pressure:</p> <ul> <li><strong>A priority label.</strong> P1 over P3. In this fixture it’s a preference ordering: which thing you’d rather do first, all else equal, and it is not a time. If policy ties each tier to a binding clock, the label carries that commitment too. Find out which kind you have before you argue from it.</li> <li><strong>A service commitment with a clock attached.</strong> “First substantive action within fifteen minutes.” Somebody promised this to somebody, and changing it takes whoever actually holds that authority. Sometimes that’s an internal service owner acting alone; sometimes it needs the agreement of the party the promise was made to, and some promises aren’t anybody’s to quietly move. In this model that authority sits above the shift.</li> <li><strong>A physical expiry.</strong> The supplementary log source rolls off in twenty-five minutes. Retention and preservation arrangements can sometimes be changed before a source ages out, by the people who administer it - but not here. In this window the expiry is a fixed input and nobody in the fixture can move it.</li> <li><strong>A declared non-waivable obligation.</strong> This one I’ve defined into the model: a requirement the lead cannot set aside whatever the queue looks like. I’m not telling you which real-world legal or regulatory duties behave this way, and you shouldn’t take one from me. I’m stipulating one so the schedule has a hard edge in it.</li> </ul> <p><br /></p> <p>The useful question isn’t which of the four outranks which - it’s what each requires, who can change it, and which constraints are fixed. A priority label alone doesn’t answer that without the policy behind it. One of the SOC managers interviewed by Kokulu et al. described their version of the second category as a ladder - <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">“In order to mitigate a P0 we have 30 minutes, P1 60 minutes, P2 120 minutes, P3 24 hours, P4 48 hours”</a> - and reported hitting it 87 percent of the time while being unhappy about it. That’s one participant’s self-reported account from an eighteen-person interview study across six industry sectors, not a measured service level and not a policy recommendation; the paper itself is careful that its results <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">may not generalize to all SOCs</a>. I’m citing it because it shows the shape - a priority label fused to a clock - not because it establishes the numbers.</p> <p>A ladder like that gives you response clocks. It doesn’t, by itself, tell you where the displaced work will fit.</p> <p><br /></p> <p><strong>What the interrupt guidance gives us</strong></p> <p>The best practitioner writing on interruption I know of is <a href="https://sre.google/sre-book/dealing-with-interrupts/">chapter 29 of Google’s SRE book</a>, by Dave O’Connor. Its focus is how a team organizes interrupt work. I’m asking a narrower question here: how to sequence accepted work when this shift’s staffing, eligibility and ownership are fixed.</p> <p>The chapter lists the metrics SRE teams found useful for deciding how to handle interrupts: expected response time, how many interrupts are usually backlogged, their severity, their frequency, and <a href="https://sre.google/sre-book/dealing-with-interrupts/">the number of people available to handle a certain kind of interrupt</a>. It also says plainly that all of those metrics are <a href="https://sre.google/sre-book/dealing-with-interrupts/">“suited to meeting the lowest possible response time, without factoring in more human costs,”</a> and taking stock of the human and productivity cost is difficult.</p> <p>Its prescriptions follow from that framing, and they’re all organizational. Polarize time so a person knows at the start of the day whether they’re on projects or on interrupts. <a href="https://sre.google/sre-book/dealing-with-interrupts/">If the volume of one class of interrupt is too high for one person, add another person</a>. Nobody should <a href="https://sre.google/sre-book/dealing-with-interrupts/">be expected to be on-call and also make progress on projects</a> or anything else with a high context-switching cost. Build a handoff for tickets as well as for on-call, because <a href="https://sre.google/sre-book/dealing-with-interrupts/">a handoff process maintains shared state as responsibility switches over</a>.</p> <p>Those are useful options when they’re available. The chapter explicitly offers <a href="https://sre.google/sre-book/dealing-with-interrupts/">components you can implement piecemeal</a> if the whole model doesn’t suit you, and one of its suggestions is that an overloaded on-call <a href="https://sre.google/sre-book/dealing-with-interrupts/">can start bumping things to their secondary, or downgrading pages to tickets</a>. A real lead with an idle secondary has a lever tonight. In the fixture I’m about to build, staffing, eligibility and ownership are held fixed, and arrivals are already accepted. I’m using that constraint set to isolate the sequencing decision, not to declare organizational changes impossible during a shift.</p> <p>The chapter also gives a context-switch estimate: <a href="https://sre.google/sre-book/dealing-with-interrupts/">“A 20-minute interruption while working on a project entails two context switches; realistically, this interruption results in a loss of a couple hours of truly productive work.”</a> That’s a manager’s working estimate about software project work, offered as part of guidance <a href="https://sre.google/sre-book/dealing-with-interrupts/">“based on what’s worked for various SRE teams”</a> at Google. It is not a measured restart penalty for an identity investigation. Using it as one would put an unsupported cost into the comparison. What does transfer is the instruction one sentence earlier: <em>assign a cost to context switches</em>. Put a number on the switch, make the number visible, and let people argue with the number instead of with each other’s vibes.</p> <p>One more borrowed caution before the example, because the rest of this post is made of minutes. Reading across their interviews, Kokulu et al. conclude that quantitative metrics such as incident counts and average response time <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">were not effective at measuring SOC success, “because each security event has unique severity and consequences”</a>; one of their manager participants put it as <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">“the metrics that just show you numbers instead of the impact are generally not so effective.”</a> That’s the authors’ qualitative reading of what eighteen interviewees told them, not an experiment showing those metrics fail - and it’s enough for the narrow use I’m making of it. Minutes are a scheduling currency. They are not a harm currency. Nothing below converts eight minutes into avoided damage, and when I say one option is better I mean better against a stated commitment, not safer.</p> <p><br /></p> <p><strong>Ninety minutes, two analysts, six cases</strong></p> <p><em>Editorial Note: the worked example is invented - the analysts, the cases, the durations, the deadlines, the delegated authority and the interruption costs are all stipulations I wrote to make the reasoning inspectable. The sourced claims interleaved with it from here on are real and carry their links; the fiction is the fixture, not the citations. I checked the schedules for internal consistency - contiguous occupancy, no double-booking, task-minute conservation, and compliance with each declared deadline - and that is the only thing the arithmetic establishes. It is not a measurement, a simulation of any real environment, or evidence about how long this work actually takes anywhere.</em></p> <p>The window runs from minute 0 to minute 90. Two analysts, both available the whole time, no meetings and no breaks inside it. That last stipulation is a modeling convenience for a bounded interval. It is not a utilization recommendation, and nothing here says a real roster should be built this way.</p> <p>Alex is cybersecurity investigator, eligible for cases I, B1 and U. Blair is an investigator and the shift lead, eligible for O, B2 and N, and also carries all routine intake. Intake is five minutes at each of minutes 0, 20, 45, 70 and 85 - twenty-five charged minutes in total, with no more than twenty-five minutes between the start of one check and the start of the next. That’s periodic coverage, not continuous monitoring. Each block covers queue review, acknowledgements, changed-risk and deadline review, and the scheduling communication that goes with them, and five minutes is enough only because this fixture contains exactly the arrivals it contains. Intake is charged time, not free supervision.</p> <p>Everything that arrives has already been accepted upstream. That acceptance is a fixed input to the model, not an uncredited third analyst doing work off the books - there is no third analyst anywhere in this window.</p> <p>Minute 90 is a reporting boundary. It is not shift change, not case closure, and not a queue reset. Work in flight at 90 is still in flight at 91.</p> <p>Every duration below is remaining <em>active</em> analyst time, with the orientation, recording and communication inside each unit stated explicitly rather than assumed away. Setup, pauses and coordination are all budgeted somewhere. Waiting on somebody else is the one thing that isn’t: it consumes elapsed time and nobody’s minutes - though the checks and coordination that go with waiting are charged like everything else. “First substantive action” means the first real move on the case after the two minutes of reading and setup, not the acknowledgement.</p> <table> <thead> <tr> <th>Case</th> <th>Owner</th> <th>On the board at</th> <th>Remaining active minutes, released unless noted</th> <th>What it owes</th> </tr> </thead> <tbody> <tr> <td>I</td> <td>Alex</td> <td>already running</td> <td>10 min checkpoint, then 40 min analysis</td> <td>status update finished by 90; no completion deadline</td> </tr> <tr> <td>O</td> <td>Blair</td> <td>accepted 120 min before 0</td> <td>15 min preservation, then 4 units of 10 min</td> <td>preservation verified by 25; remaining handling by 105</td> </tr> <tr> <td>B1</td> <td>Alex</td> <td>0</td> <td>25 min</td> <td>first substantive action by 15; package by 40</td> </tr> <tr> <td>B2</td> <td>Blair</td> <td>0</td> <td>20 min</td> <td>first substantive action by 30; package by 50</td> </tr> <tr> <td>U</td> <td>Alex</td> <td>40</td> <td>10 min containment, plus 30 min of follow-up blocked until 110</td> <td>first action by 44; containment verified by 54</td> </tr> <tr> <td>N</td> <td>Blair</td> <td>60, read during intake at 70</td> <td>10 min, then 15 min</td> <td>first action by 105; package by 130</td> </tr> </tbody> </table> <p><em>The case-minute totals above exclude separately scheduled intake and reporting. I’s two-minute status update is counted as its own obligation, as are Blair’s five-minute intake blocks. Pause, state capture and re-entry costs are also accounted for separately in the schedules below.</em></p> <p>A few case details matter later. I already has twenty minutes of work behind it: a known account-access path is contained and the initial records are saved, but related activity is unresolved and Alex is partway through building the next evidence map. O has a supplementary evidence source that expires at minute 25, and its ten-minute collection segment can’t be preempted once started. U reports ongoing unauthorized activity and triggers a pre-authorized immediate-action requirement; its containment action and verification can complete independently of the full investigation, which is a separable unit of work, not a closed case. N arrives with a temporary precaution already in place, which is a reason its handling may wait and not evidence that it’s harmless.</p> <p>The unit costs matter later too, so here they are in full. I’s checkpoint is eight minutes of work plus two of recording; its analysis is four ten-minute units, each including the recording of its own result. O’s preservation is two minutes of setup, ten of collection that can’t be preempted once started, and three of verification and recording; each of O’s four later ten-minute units is two minutes of orientation, six of substantive work and two of recording and communication. B1 is two minutes of reading and setup, twenty of substantive work and three of completion and communication. B2 has the same shape at two, fifteen and three. U’s containment is two of setup, six of action and two of verification and recording, and its final eight minutes need uninterrupted attention. N’s first unit is two, six and two; its finishing unit is two minutes of reorientation, ten of substantive work and three of reporting. Nothing is hiding in a rounding error.</p> <p>Before minute 0, the service owner authorized bounded sequencing inside the declared deadlines, including protected return steps for displaced work. No case changes priority or owner between the options below. Higher ordinary priority is a preference in this model rather than an unconditional preemption rule - the expiry, the mandatory action deadline and the intake commitments all constrain it. The lead cannot waive the physical expiry or the declared non-waivable obligation, and nothing in this window creates a new notification duty.</p> <p>Now the arithmetic that looks damning and isn’t. Add up the remaining case work released before minute 90: I needs 50 minutes, O needs 55, B1 needs 25, B2 needs 20, U’s containment needs 10, and N needs 25. That’s 185 case-minutes. Blair also has 25 minutes of intake, and I has a separate two-minute status update due by 90. The total is 212 minutes against 180 available analyst-minutes. That is a minimum shortfall of 32 minutes before accounting for any pause or re-entry overhead. Another thirty minutes of U’s follow-up is sitting outside the window entirely, blocked on an external result.</p> <p>That proves exactly one thing, and it’s worth having: under these inputs, all of the released work and its associated commitments cannot be completed by minute 90. Something remains unfinished at the horizon no matter what the lead does, and the only live question is which something.</p> <p>What it does not prove is that anything is late, because not all of that work is due by 90. O’s remaining handling is owed at 105. N’s package is owed at 130. I has no completion deadline at all, only a status update. So “released work exceeds available minutes” is not a breach, not proof that the commitments are mutually incompatible, not evidence of an unsustainable arrival rate, and not a staffing case. Here it’s just what a queue with a horizon in it looks like. The sharper finding arrives when the work that must <em>finish</em> inside a span exceeds the eligible minutes inside that span, and we’re going to find exactly one place where that happens.</p> <p><br /></p> <p><strong>Minute zero: the choice that looks like it’s about focus</strong></p> <p>At minute 0 Alex is mid-unit on I and two qualified burst cases land, B1 to Alex and B2 to Blair. Blair’s side is uncomplicated and identical under both options: intake 0 to 5, O’s preservation 5 to 20 so the expiring source is captured with five minutes to spare, intake 20 to 25, B2 from 25 to 45, intake 45 to 50, then O’s three review units at 50 to 60, 60 to 70 and 75 to 85 with the minute-70 intake block sitting between them, and the last intake block from 85 to 90. Blair’s first substantive action on B2 lands at 27 against a deadline of 30, and B2’s package closes at 45 against 50. Tight, inside every declared deadline, and unaffected by what Alex decides.</p> <p>Alex has an actual decision. Option one: stop I now, pay the two-minute minimum state capture, and start B1. Option two: finish I’s ten-minute checkpoint first, then start B1.</p> <p><br /></p> <p>Option one, minute by minute:</p> <table> <thead> <tr> <th>Minutes</th> <th>Alex</th> </tr> </thead> <tbody> <tr> <td>0-2</td> <td>pause I, capture current state</td> </tr> <tr> <td>2-27</td> <td>B1, start to finished package</td> </tr> <tr> <td>27-33</td> <td>re-enter the interrupted I unit</td> </tr> <tr> <td>33-40</td> <td>I checkpoint, 7 of 10 minutes</td> </tr> <tr> <td>40-42</td> <td>pause I again</td> </tr> <tr> <td>42-52</td> <td>U containment and verification</td> </tr> <tr> <td>52-58</td> <td>re-enter the interrupted I unit</td> </tr> <tr> <td>58-61</td> <td>finish the I checkpoint</td> </tr> <tr> <td>61-88</td> <td>I analysis</td> </tr> <tr> <td>88-90</td> <td>I status update</td> </tr> </tbody> </table> <p><br /></p> <p>Option two:</p> <table> <thead> <tr> <th>Minutes</th> <th>Alex</th> </tr> </thead> <tbody> <tr> <td>0-10</td> <td>I checkpoint, complete</td> </tr> <tr> <td>10-35</td> <td>B1, start to finished package</td> </tr> <tr> <td>35-37</td> <td>reorient from a saved checkpoint</td> </tr> <tr> <td>37-40</td> <td>I analysis</td> </tr> <tr> <td>40-42</td> <td>pause I</td> </tr> <tr> <td>42-52</td> <td>U containment and verification</td> </tr> <tr> <td>52-58</td> <td>re-enter the interrupted I unit</td> </tr> <tr> <td>58-88</td> <td>I analysis</td> </tr> <tr> <td>88-90</td> <td>I status update</td> </tr> </tbody> </table> <p><br /></p> <p>Both fill exactly ninety minutes with no idle time. Both satisfy every declared in-window requirement: B1’s first action, B1’s package, O’s preservation, B2’s first action and package, U’s first action and verified containment. Nothing is late in either.</p> <p><br /></p> <p><img width="1600" height="1200" alt="A ninety-minute timeline comparing two alternative schedules for Alex alongside Blair’s unchanged schedule. Both Alex paths interrupt work at minute 40 to handle U containment, while Blair handles preservation, B2, O reviews, and routine intake." src="https://github.com/user-attachments/assets/64d5f47d-332f-4a34-b621-9bd45e38cd48" /></p> <p><br /></p> <p>So what actually differs?</p> <table> <thead> <tr> <th> </th> <th>Take B1 immediately</th> <th>Finish the checkpoint first</th> </tr> </thead> <tbody> <tr> <td>B1 first substantive action</td> <td>minute 4</td> <td>minute 12</td> </tr> <tr> <td>B1 package complete</td> <td>minute 27</td> <td>minute 35</td> </tr> <tr> <td>I case-minutes completed</td> <td>37</td> <td>43</td> </tr> <tr> <td>I minutes outstanding at 90</td> <td>13</td> <td>7</td> </tr> <tr> <td>Separate I costs charged, including the scheduled status update</td> <td>18 min</td> <td>12 min</td> </tr> <tr> <td>Pause and re-entry alone</td> <td>16 min</td> <td>10 min</td> </tr> <tr> <td>Loaded case-minutes worked across all cases</td> <td>137</td> <td>143</td> </tr> <tr> <td>Intake minutes</td> <td>25</td> <td>25</td> </tr> <tr> <td>Total occupancy</td> <td>180</td> <td>180</td> </tr> <tr> <td>Ready case-minutes left across all cases at 90</td> <td>48</td> <td>42</td> </tr> <tr> <td>Blocked minutes left at 90</td> <td>30</td> <td>30</td> </tr> </tbody> </table> <p><br /></p> <p>That’s the whole trade. Eight minutes earlier on B1, against six more minutes of real progress on I. O, B2, U and N are untouched, intake is untouched, and both analysts are busy for all ninety minutes either way - but equal occupancy is not equal case work. The checkpoint option converts six minutes of pause and re-entry into six minutes of actual case work, which is why loaded case time comes out at 143 rather than 137 against the same 180 minutes of occupancy. Note too that the 18 and 12 include the two-minute status update at 88, which is scheduled reporting rather than an interruption cost; pause and re-entry on their own are 16 and 10.</p> <p>Two things about those six minutes matter before calling them a benefit of “protecting flow.” First, they’re not productivity, focus quality or recovered concentration. They decompose into the two-minute partial pause that option two never pays, plus four minutes of difference between re-entering a half-finished unit from scratch (six minutes) and reorienting from a completed checkpoint (two). Those are stipulated costs. Change them and the advantage changes or vanishes. Second, six minutes of analysis is not six minutes of safety. I can tell you option two leaves seven minutes of declared work outstanding instead of thirteen. I cannot tell you it leaves the organization in a better position, and I’m not going to pretend otherwise.</p> <p>The same distinction applies to B1. Eight minutes earlier is a real difference in timing, measured off the same stipulated clock. What those eight minutes are worth in avoided harm is exactly as unknown as what the six on I’s side are worth. Neither converts into security value here, and quietly treating the unquantified side of a trade as the zero side is the move I am trying to avoid.</p> <p>There’s also a hard bound on the checkpoint choice. B1’s first substantive action is the checkpoint length plus two minutes of setup. Its fifteen-minute start deadline therefore requires a checkpoint of thirteen minutes or less. The nominal estimate is ten, so there’s three minutes of room; beyond thirteen, the checkpoint-first option violates B1’s start deadline. That’s a local necessary condition, not a revalidation of some altered schedule; stretch the checkpoint and you have to re-check everything downstream, not just that one deadline.</p> <p>My own judgment, stated as a judgment: with a service owner who has pre-authorized bounded sequencing, with three minutes of slack on the binding constraint, and with nothing in B1’s packet suggesting it is more urgent than its own deadlines say, I’d take the checkpoint. Not because uninterrupted work is sacred, but because the cost of the alternative is concrete and paid twice while the benefit of the alternative - eight minutes earlier on a case with a forty-minute package deadline that both options meet - is unquantified. Unquantified is not zero, and if somebody can tell me what those eight minutes are actually worth on this case, that’s the argument that moves me. Flip the authorization, remove the slack, or give B1 a reason to be fast that isn’t just its label, and I’d flip with it.</p> <p><br /></p> <p><strong>Minute forty: why this isn’t an argument for uninterrupted work</strong></p> <p>At minute 40 a separate qualified account incident arrives on the emergency channel reporting ongoing unauthorized activity. U invokes the pre-authorized immediate-action requirement: first substantive action by minute 44, containment verified by 54.</p> <p>Here’s where the “protect the analyst’s focus” argument has to earn its keep, and doesn’t.</p> <p>In option one, Alex is three minutes short of finishing the checkpoint unit. Letting it finish puts U’s first action at minute 45. In option two, Alex is three minutes into a ten-minute analysis unit; letting it finish puts U’s first action at 49. The deadline is 44. Both are late. Under these inputs there is no version of “let them finish the thing they’re doing” that works, so both trajectories interrupt, both pay the two-minute pause, and both run U from 42 to 52 - first action at 44 with exactly zero slack, verified finish at 52 with two minutes to spare against 54.</p> <p>To me, that’s the most important point here: the interruption at minute 40 costs real capacity and is still required. It costs Alex eight minutes of overhead across the pause and re-entry. Waiting for the current unit to finish would miss U’s first-action deadline; interrupting now lets Alex take and verify the specified action inside both mandatory deadlines. That establishes a specified action against activity that was ongoing, plus its record. It doesn’t erase that activity or tell us what may have happened elsewhere. This is a claim about an obligation, not about which eight minutes were most valuable. Any argument about interruption has to account for both the cost and the obligation.</p> <p>Notice what made that possible: the containment action is a <em>unit</em>, not a case. FIRST describes exactly this kind of work as <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">short-term tactical action to stop immediate damage and limit the extent of malicious activity</a>, separate from the analysis and the recovery. Ten minutes buys a specified action and its verification record. It does not buy eradication, root cause, or a closed case.</p> <p>And U’s remaining thirty minutes of follow-up can’t start before minute 110, because it depends on an external evidence result that isn’t available until then - and might not be available then either. That waiting consumes elapsed time, not analyst-minutes, which is why those thirty minutes sit outside the ninety-minute capacity arithmetic. The checking doesn’t get a free pass, though: somebody has to look at minute 110, and that look is charged like anything else. FIRST has a name for keeping track of this: the detailed-analysis-coordination function exists to initiate and track work done by others, producing <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">a list of pending and outsourced analysis</a> from the handler’s point of view. Kokulu’s participants described the operational version of the same thing - one manager whose SOC only detects, hands cases to other teams and <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">waits for their completion, which renders some metrics inaccurate</a>, and an analyst in an outsourced SOC who said <a href="https://yancomm.net/papers/2019%20-%20CCS%20-%20Matched%20and%20Mismatched%20SOCs.html">“we do not do the final actions to remediate so we cannot be very fast.”</a> Those are self-reported accounts of dependency, not evidence that waiting or chasing is free. So a blocked case needs two things rather than one: a review time for the dependency, and eligible capacity for the work itself once the dependency clears. Nothing prohibits reserving a conditional block in advance either - it’s just conditional, which means you need the review either way.</p> <p><br /></p> <p><strong>What a pause actually holds</strong></p> <p>The pause costs are small and they’re doing a specific job. Two minutes captures the current evidence state and the cursor, and notifies whoever needs to know that the case stopped. Six minutes reconstructs and re-checks that context before real progress resumes. Two minutes reorients from a completed checkpoint, which is cheaper because the unit boundary did the work in advance.</p> <p>What those minutes buy is retained state for an obligation that still exists. They don’t settle any of these separate questions:</p> <ul> <li><strong>A pause is not closure.</strong> I is in progress at 90 in both options, with its related activity still unresolved. The containment that predates the window addressed one known account-access path. It doesn’t make further delay harmless, and it doesn’t tell you the case is safe to leave - the unexamined related activity is precisely the part nobody can characterize yet.</li> <li><strong>A pause is not a transfer.</strong> Nobody else became eligible. Alex still owns it.</li> <li><strong>A pause is not a priority change.</strong> N keeps the higher ordinary priority it arrived with even while it sits behind O, because the reason it waits is an authorized window, not a demotion.</li> <li><strong>A pause is not authorization.</strong> FIRST’s glossary borrows a project-management definition of risk acceptance: a strategy where you <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">acknowledge the risk and take no action unless it occurs</a>. Read that for what it is - one glossary’s borrowed wording, not a taxonomy that makes deferral and acceptance mutually exclusive. They aren’t. Committing to act later and accepting whatever exposure accumulates while you wait are two separate decisions, and the exposure includes activity already underway. Sequencing is the first decision. The second needs somebody with the standing to accept it, and a pause doesn’t confer that standing any more than a tidy pause note confers safety.</li> </ul> <p><br /></p> <p>And the distinction I’d push hardest on: a record does not create capacity. Writing a beautiful pause note with retained state, next step and risk commentary doesn’t reserve a minute to resume the case. It makes the obligation legible to other people, which is the job the SRE chapter describes when it says a handoff <a href="https://sre.google/sre-book/dealing-with-interrupts/">maintains shared state between ticket handlers</a>. But the minutes to resume still have to come from somewhere, and the queue does not notice that you documented it.</p> <p>An unwritten plan can still be a plan. If the lead knows the next step, the step is feasible, and the person who’d execute it knows too, they’ve identified a feasible return. A tidy note with no identified time doesn’t establish that. That’s a comparison of what we know about resumption, not a verdict on which shift is better run. Writing the plan down makes it shareable; it doesn’t create the time. Where a record is required it is required, and “I had it in my head” is not a defense.</p> <p><br /></p> <p><strong>The return trip comes out of somebody’s hour</strong></p> <p>So let’s make the lead pay for one. Here’s the conditional continuation on Blair’s side after minute 90 - conditional because it depends on the same people staying available, the estimates holding, the reported precautions remaining applicable, and no new non-deferrable demand arriving.</p> <p><br /></p> <table> <thead> <tr> <th>Minutes</th> <th>Blair</th> <th>Against</th> </tr> </thead> <tbody> <tr> <td>90-100</td> <td>O’s final decision unit</td> <td>O’s remaining handling owed at 105</td> </tr> <tr> <td>100-110</td> <td>N’s first unit, first action at 102</td> <td>N’s first action owed at 105</td> </tr> <tr> <td>110-115</td> <td>intake</td> <td>check spacing stays within 25 minutes</td> </tr> <tr> <td>115-130</td> <td>N’s finishing unit</td> <td>N’s package owed at 130</td> </tr> </tbody> </table> <p><br /></p> <p>Everything lands. O finishes five minutes early, N starts three minutes early, N’s package closes exactly on its deadline.</p> <p>O gets a reserved, named, ten-minute block of Blair’s time even though N arrived at minute 60 and carries higher ordinary priority. Not “we’ll pick it back up when things calm down.” A person, a slot, and a reason: O’s commitment is approaching at 105, N’s authorized waiting window extends to 105 and 130, and both fit. Blair also doesn’t discover this at minute 90 - the 70-to-75 intake block is where O’s approaching deadline gets reviewed and its remaining units get reserved, with confirmation during 85 to 90. That’s part of what the intake budget pays for.</p> <p>Now count the continuation. Blair has forty minutes between 90 and 130, and the plan uses ten for O, twenty-five for N and five for intake. Forty out of forty. There is no spare Blair minute in that span.</p> <p><br /></p> <p><img width="1200" height="896" alt="Infographic timeline tracking a 40-minute span across four blocks: 90–100 (O's final decision unit, 10 min), 100–110, 110–115 (Intake, 5 min), and 115–130 (N's finishing unit, 15 min). A bracket underneath indicates that all 40 out of 40 available minutes are committed with zero spare capacity. A separate red box labeled points to a stop icon with the text Nowhere to go before minute 130." src="https://github.com/user-attachments/assets/c555f4ac-11aa-4cc6-a9a4-b6fd7c819fa1" /></p> <p><br /></p> <p>That’s the honest cost of the promise, and it’s the part a reassuring “don’t worry, it’s reserved” leaves out. Reserving O’s decision unit didn’t conjure time. It spent time that N could have used. N still meets both of its declared deadlines, which is a statement about clocks rather than about N being fine - nothing in this fixture tells me what N’s extra waiting costs. Returning to older work is not free and it is not virtuous; it’s a choice that displaces something else. Here, I can show that the displaced work still fits under the delegated sequencing authority. That supports the scheduling choice; it doesn’t settle whether the exposure while N waits is acceptable.</p> <p>FIRST’s description of ordinary response planning says this more politely than I will: determine a plan that provides a meaningful response <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">“within the desired business requirements and timeframe based on available resources and the technical scope of required actions.”</a> Based on available resources. The framework also expects that <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">whenever new analysis results become available, the new findings need to be reviewed</a>, which is one way a reservation legitimately dies: something changed about O itself.</p> <p>It isn’t the only way. A genuine emergency on a different case can take Blair’s 90-to-100 block too, and that override has nothing to do with O’s evidence at all. Either way the reservation fails, and what you owe when it fails is specific. Name the commitment now at risk - here, O’s remaining handling at 105. Say who still owns the case, what state its evidence is in, and how many minutes of work are still outstanding. Say who looks at it next and when. Then put it in front of whoever can actually change the commitment, if nobody can make it fit. All of that is charged time; the communication and the escalation come out of the same forty minutes as the work. And escalating is not the same as being rescued: an answer may not arrive in time, and the authority to move O’s 105 might not be reachable in the minutes you have.</p> <p><br /></p> <p><strong>Where it actually breaks</strong></p> <p>So far, the declared commitments fit. Now change one input - and note that this is a change to the fixture, not something the earlier arithmetic already contained. Suppose an additional obligation lands on Blair inside the 90-to-130 span: a case only Blair can touch, requiring a strictly positive amount of active work, and <em>due</em> inside that same span. Not an ordinary arrival that can wait under the same authority. Arriving in the window isn’t enough. It has to be owed in the window.</p> <p>Now there is nowhere to put it. Blair’s forty minutes are already spoken for - ten for O’s decision, twenty-five for N, five for the intake block at 110 - and that intake block is a commitment too, not slack wearing a label. Alex isn’t eligible. There is no third analyst. So at least one obligation now fails: O’s handling at 105, one of N’s two clocks, the intake check, or the new work itself. Which one fails is a decision somebody has to make, and making it needs either the authority to change a commitment or capacity that doesn’t currently exist. It is not a sequencing problem the lead can solve by being clever.</p> <p>Released work exceeding available minutes - 185 against 180, items carried past a reporting boundary, a case sitting unworked for an hour - is not by itself a failed service. Here, the capacity witness is stronger: once the additional obligation is included, the work that must finish inside the 90-to-130 span exceeds the eligible minutes inside it. That’s sufficient to establish infeasibility; it isn’t a definition covering every way a schedule can become impossible. Release times, eligibility, precedence and non-preemptible windows can each rule out an otherwise plausible plan. If the only eligible person is midway through a collection segment that can’t be stopped before the deadline passes, the raw minutes can add up perfectly well and the plan is still dead.</p> <p>In this fixture, the operational difference is between carrying unfinished work and escalating incompatible commitments.</p> <p>Escalation asks someone with the authority to change a commitment to decide how to resolve the conflict. FIRST’s reporting function describes the output as internal stakeholders being apprised of current activities, completed actions and pending ones, with <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">the assessed impact of delays, recommendations and requested actions</a> communicated alongside. That’s the artifact: here is what’s committed, here is what cannot all happen, here is what I recommend giving up and what it costs.</p> <p>The request doesn’t suspend existing duties or guarantee an answer in time. It doesn’t move a physical expiry or make a declared non-waivable obligation waivable. In this model the authority to change a service commitment sits above the shift - a property I chose deliberately, because it’s the version that makes the decision hard. FIRST’s most explicit statement about postponing a response sits in its crisis-management section, where <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">a high-ranking role takes over and changes the usual line of command for the duration of the emergency</a>, and where the crisis team <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">“may decide to postpone the response to an actual information security incident.”</a> The same section notes that a decision to discontinue a response or carry on <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">can only be taken based on the best situational awareness available</a>, and that when such a decision is made, <a href="https://www.first.org/standards/frameworks/csirts/csirt_services_framework_v2.1">it needs to be communicated to all entities currently informed and participating</a>. That’s a formal crisis mechanism, and it isn’t the only legitimate route by which a commitment ever changes - the framework’s ordinary planning and reporting functions already assume finite resources and communicated delays. What it isn’t is a routine shift-level waiver. It’s a documented transfer of authority under declared emergency conditions, not permission for this lead to waive the ordinary commitments.</p> <p>So the realistic list is short. A lead in this position can promise a named owner and an explicit retained state, a reserved next step or paid dependency review under stated conditions, honest recognition and communication of a threatened commitment, and escalation within whatever delegation actually exists. A lead cannot promise that nothing bad happens while work waits, completion in the face of unbounded arrivals, capacity produced by relabeling something, delivery of an external result, or authority the organization never handed over.</p> <p><br /></p> <p><strong>Past minute ninety</strong></p> <p>Nothing is clean at the horizon, which is the point of putting the horizon there.</p> <p>I is still open, with seven or thirteen minutes of declared work left depending on which option the lead took at minute zero, and a status update at 88 that just told somebody precisely which. Its related activity is still unresolved. Alex carries straight on with the same unit after 90 - a reporting boundary isn’t an interruption, so no fresh re-entry gets charged - and on the nominal estimates that declared work lands at 97 or 103, conditional on Alex staying available and nothing authorized displacing it. “Declared work finished” is not “case closed.”</p> <p>O has one ten-minute decision unit and a slot on Blair’s calendar at 90 that exists because someone reserved it at minute 70 and confirmed it at 85.</p> <p>U has a verified containment record from minute 52 and thirty minutes of follow-up that cannot start before 110. Somebody has to check at 110 whether the external result actually arrived, and I’ll stipulate what that check costs rather than leave it free: two minutes of Alex’s time, available because Alex’s declared I work is done well before then. If the result landed and Alex remains available, the thirty minutes can begin after that check - which gets the follow-up moving and closes nothing, since this fixture gives it no finish deadline at all. If the result didn’t land, what U gets is a paid coordination step and another conditional review time. Not an open-ended promise, and not unbounded free chasing.</p> <p>N has been read during the minute-70 intake block, acknowledged, and parked with a start and a finish attached, and its reported precaution is still a precaution rather than a finding. B1 and B2 have finished their declared handling packages - which is narrower than “nothing more will ever come of them,” and is still the only part of this window that looks finished.</p> <p>That’s the shape of a hard shift that meets its declared commitments. Not an empty queue - under these inputs an empty queue was never available, and any version of this story that produces one has quietly added a third analyst, waived a commitment, or pretended the reporting boundary wipes the board. What this shift produces is a set of displaced obligations you can still describe accurately.</p> <p>Which gives the questions I’d actually ask, at minute 90 or at handoff or in a postmortem three weeks later. For every item still waiting on work: who owns it, what state is it in, whose time is it in next, and what would take that time away? Four questions.</p> <p>Be clear about what answering them does. It makes the decision inspectable - by the next shift, by the service owner, by the person whose case waited. It does not establish that the lead had the authority to make that call, that the exposure piling up behind it is acceptable, that the plan is feasible, or that the outcome will be good. Those judgments need their own grounds, under the applicable delegation. The four answers show what was promised and what it depends on.</p> <p>And the third one has a legitimate hard case. Work blocked on somebody else’s evidence may genuinely not have a restart slot yet, because the restart depends on something nobody in the building controls. The honest answer there is a review time and the condition attached to it, and that is a real answer. “When things calm down” isn’t a lie or proof of abandonment. On its own, though, it leaves the person, time and usable condition behind the promised return unspecified. That leaves too little to inspect. If that’s the best answer you have for a case, the useful move is to go and find the three missing pieces.</p> <p><img src="http://canarytokens.com/traffic/wa3d4w9t36ttx2alvw5n4jxht/index.html" style="display: none;" /></p>
  108. Srsly Risky Biz: The Ritual Bitcoin Sacrifices Will Continue Until Security Improves

    Thu, 05 Nov 2026 04:05:57 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://sublime.security/"><em><u>Sublime Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/2026-11-05-3840px-Hubert_van_Eyck_004.jpg" alt="Srsly Risky Biz: The Ritual Bitcoin Sacrifices Will Continue Until Security Improves"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://sublime.security/"><em><u>Sublime Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB189.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB189/&quot;&gt;Srsly Risky Biz: The ritual bitcoin sacrifices will continue until security improves&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 17:43 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: The ritual bitcoin sacrifices will continue until security improves&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-2155c4fd-72a3-4d17-a8c7-8b99d13566bf.jpeg" class="kg-image" alt="Srsly Risky Biz: The Ritual Bitcoin Sacrifices Will Continue Until Security Improves" loading="lazy" width="2000" height="1125" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/11/data-src-image-2155c4fd-72a3-4d17-a8c7-8b99d13566bf.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/11/data-src-image-2155c4fd-72a3-4d17-a8c7-8b99d13566bf.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/11/data-src-image-2155c4fd-72a3-4d17-a8c7-8b99d13566bf.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/11/data-src-image-2155c4fd-72a3-4d17-a8c7-8b99d13566bf.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Hubert and Jan van Eyck, Adoration of the Mystic Lamb</span></figcaption></figure><p>North Korea has pillaged the Binance cryptocurrency exchange to the tune of USD&#x24;6.3 billion. With results like these, it&apos;s no wonder the country&apos;s operators have given up on targeting real, actual banks.</p><p>In last Thursday&apos;s heist, tokens from 11 different cryptocurrencies were stolen. This is the largest theft in history, dwarfing the <a href="https://www.elliptic.co/insights/bybit-hack-largest-in-history/"><u>&#x24;1.46 billion Bybit hack</u></a> in February last year.&#xa0;</p><p>Binance was able to get about &#x24;2.8 billion worth of the stolen <a href="https://en.wikipedia.org/wiki/Stablecoin"><u>stablecoins</u></a> frozen, leaving &#x24;3.25 billion in currencies that cannot be frozen, mostly in Bitcoin and Ethereum. Less than &#x24;6.3 billion, sure. But still huge. Blockchain analysis companies <a href="https://www.chainalysis.com/blog/binance-exploit-latest-findings-and-response/"><u>pretty quickly</u></a> linked the hack to North Korea and by the weekend the FBI <a href="https://www.fbi.gov/investigate/cyber/alerts/2026/north-korea-responsible-for-6-billion-binance-attack"><u>attributed the hack</u></a> to a North Korean group it calls TraderTraitor.&#xa0;&#xa0;</p><p>Binance&apos;s <a href="https://x.com/binance/status/2117626126214504729"><u>initial incident response reports</u></a> what it calls a &quot;sophisticated multi-stage campaign&quot;. TraderTraitor made its first inroads into Binance in June this year when &quot;engineer A&quot; installed a <a href="https://isc.sans.edu/diary/32942"><u>malicious version</u></a> of <a href="https://en.wikipedia.org/wiki/Homebrew_(package_manager)"><u>Homebrew</u></a> after clicking on a malvertising link. Binance&apos;s security team detected and cleaned up the malware, but TraderTraitor learned enough about the company&apos;s engineering org chart to target them in subsequent operations.</p><p>A different Binance engineer, &quot;B&quot;, was then lured into a fake job interview in which TraderTraitor used a malicious coding test to install malware on B&apos;s computer. This was a two-stage &apos;interview&apos; process, and Binance says that the first round involved a live video interview. The purported HR representative built trust in this interview by sharing a social media account that included personal photos and videos. It was only in the second interview that B was told to install the malicious coding test. Binance says this entire approach &quot;was more advanced and deceptive than previously seen social engineering approaches&quot;.</p><p>After compromising B&apos;s computer, TraderTraitor &quot;used an internal AI tool to gain control over the Wallet DevOps Engineer A and another Wallet Software Engineer C&quot;. We assume they mean the engineers&apos; devices, but, come again? This feels like an important part of the &#x201c;sophisticated multi-stage campaign&#x201d;.</p><p>Regardless, this access granted TraderTraitor all the authority it needed to make off with over &#x24;6 billion.&#xa0;</p><p><a href="https://x.com/AltcoinDaily/status/2117683261694755024"><u>Speaking to <em>CoinDesk</em></u></a>, Binance founder Changpeng Zhao, who is better known as CZ, said the incident was a &quot;real wake up call, not just for crypto, but for the financial system overall&quot;.&#xa0;</p><p>&quot;If a crypto exchange can be hacked, banks are in a worse position in my opinion&quot;.&#xa0;</p><p>On a macro level, this is utter nonsense.</p><p>From a <em>narrow </em>security perspective, CZ is <em>possibly</em> right? Binance says that prior to this hack it had repelled 57 North Korea-linked attempts to compromise its systems in 2026 alone. We suspect a lot of traditional banks would have succumbed to some of those attempts. But then what? Once large amounts of stolen money start moving around the banking system, transactions start being reversed and the attackers evicted.&#xa0;</p><p>Back in 2016 North Korean actors stole USD&#x24;1 billion from Bangladesh&apos;s central bank. In that attack the banks that received the stolen money cooperated with Bangladesh Bank to recover around 90 percent of the stolen money. So North Korea only got away with around 10 percent of the stolen value vs around 50 percent for last week&apos;s cryptocurrency heist. It&apos;s not the susceptibility to hacking that makes North Korea focus on cryptocurrency, it&apos;s the ease of moving the money after the fact. That&apos;s what makes CZ&apos;s statement so... annoying?</p><p>We don&apos;t put too much hope in crypto recovery procedures reaching parity with those found in the banking sector. Sure, there are mechanisms such as sanctions that are intended to make laundering cryptocurrency more difficult, but at this point North Korea has a <a href="https://www.elliptic.co/insights/the-binance-hack-is-laundering-on-autopilot/"><u>well-oiled, automated cryptocurrency laundering machine</u></a>.&#xa0;</p><p>Which, in a way, is good news for anyone who <em>isn&apos;t</em> a crypto investor. Binance and its ilk are acting as a sort of sacrificial cash pile for the mainstream finance world. As long as the crypto world keeps absorbing the financial impact, the financial system writ large should remain safe and stable.</p><p>Crypto bros, we thank you for your sacrifice.</p><h2 id="i-feel-the-need-the-need-for-speed">I Feel the Need, the Need For Speed</h2><p>Tech executives everywhere are under immense pressure to speed up the pace of their software development with AI. From a security perspective, the results of this faster pace are going to be mixed.&#xa0;</p><p>As a case in point, this week <em>404Media</em> <a href="https://www.404media.co/manifestos-fear-and-loser-thinking-inside-protons-ai-reckoning/"><u>wrote about</u></a> privacy-focused company Proton&#x2019;s leadership-led drive to adopt AI. According to the article, an employee&apos;s request to security test newly generated code before deploying it was dismissed by CEO Andy Yen as &quot;loser thinking&quot;.</p><p>&quot;Of course it&apos;s possible to be fast and secure&quot;, Yen wrote in a Slack message seen by <em>404Media</em>. &quot;Stop thinking always of how to be slower, and focus more on how to be faster.&quot;</p><p>It is particularly jarring to hear Proton&apos;s management using language that appears to prioritise speed over security. Its first product was the end-to-end encrypted mail service ProtonMail and it has built its reputation on being a privacy and security-focussed company.&#xa0;</p><p>An internal document obtained by <em>404Media</em> indicated that the company&apos;s rush to incorporate AI was motivated by the need to move faster and keep up with the competition. The company was likely to miss its annual revenue target &quot;due to insufficient velocity&quot; and &quot;by all metrics that matter, we are an underperforming company&quot;.&#xa0;&#xa0;</p><p>As it turns out, being a privacy and security-focussed company doesn&apos;t make Proton immune from the laws of commerce. The pressure to adopt AI to move fast to keep up with competitors is being felt everywhere.&#xa0;</p><p>Michael Terkowitz, VP and and Distinguished Engineer for Worldwide Amazon Stores, acknowledged this dynamic <a href="https://www.aboutamazon.com/news/company-news/amazon-software-engineer-role-guideline-ai"><u>in a memo last week</u></a> that addressed the impact of AI on software development engineering.&#xa0;</p><p>Terkowitz said there was an imperative to adopt AI, and that &quot;We all need to embrace this change and take the time to learn, or we risk being left behind&quot;. At the same time, however, he noted that it was still early days in terms of AI deployment and that &quot;it is difficult to define&#x2026; what good AI powered engineering practices look like&quot;.</p><p>So everyone wants to use AI to write code faster, but right now not even Amazon knows how to do that properly.&#xa0;</p><p>Once it <em>does</em> zero in on the best way to use AI, we imagine Amazon will be fine from a security perspective. It has existing processes in place to prioritise it and a management structure that means security is taken seriously.&#xa0;</p><p>Its security teams report directly to the CEO and its <a href="https://aws.amazon.com/blogs/security/how-aws-built-the-security-guardians-program-a-mechanism-to-distribute-security-ownership/"><u>Security Guardians program</u></a> parachuting security champions into the company&apos;s product teams. An organisation with that kind of culture is going to do a good job of safely adopting AI. The fact that it&apos;s taking its time to figure out what the best AI-powered engineering practices look like is already a good sign.</p><p>Other tech giants are already incorporating AI into their operations to make their products more secure. Earlier this year Google&apos;s Chrome team <a href="https://blog.google/security/chrome-stronger-with-every-update/"><u>wrote about</u></a> how it had incorporated agentic workflows to improve vulnerability discovery, triage and patching.</p><p>So there is the potential for businesses to adopt software workflows that result in a more secure world. We can get behind that.</p><p>We also think tech firms should be incentivised to secure their code, perhaps by reinvigorating the US government&apos;s <a href="https://news.risky.biz/srsly-risky-biz-data-theft-extortion-is-booming-hooray/#secure-by-design-deserves-a-resurrection"><u>Secure by Design Initiative</u></a>. There&apos;s probably some room to <a href="https://therecord.media/uk-budget-proposal-software-companies-liable-cyber-breaches"><u>adjust the extent</u></a> to which softwaremakers can absolve themselves of all liability in the event of security failures, too.&#xa0;</p><p>We think the main lesson here is a simple one, though. The companies that suck at security will continue to suck at security, but faster. And the companies that take this stuff seriously will probably do well.&#xa0;</p><p>So if you&apos;re looking to AI to make bad software good, you&apos;re going to be disappointed.</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/RJb2aVDj4uM?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: The ritual bitcoin sacrifices will continue until security improves"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Google dismantles XCPE botnet:</strong> Google&apos;s Threat Intelligence Group <a href="https://cloud.google.com/blog/topics/threat-intelligence/shutting-the-door-on-the-xcpe-botnet"><u>announced</u></a> that it had dismantled a botnet consisting of more than 1.3 million Android set-top TV boxes with help from industry partners. GTIG thinks that Videostrong, a certified Android manufacturer, was deliberately adding malware to its devices. The backdoor was used for a number of malicious purposes including ad fraud and as part of a residential proxy network. Google has cut ties with Videostrong.&#xa0;</li><li><strong>Edge blocks malicious site notifications: </strong>Microsoft <a href="https://blogs.windows.com/msedgedev/2026/11/02/client-side-threat-detection-how-scareware-blocker-uncovered-malicious-notification-abuse/"><u>has announced</u></a> that its Edge browser will block &apos;scareware&apos; notifications from malicious sites. It says that scammers use these notifications to send fake virus warnings and security alerts to users.&#xa0;&#xa0;</li><li><strong>European scam infrastructure disrupted:</strong> Europol <a href="https://www.europol.europa.eu/media-press/newsroom/news/investment-scams-potential-victims-in-least-50-countries-over-100-servers-seized"><u>announced</u></a> that an international operation had arrested seven suspects allegedly involved in investment scams. Authorities searched 17 locations and seized more than 100 servers. The operation targeted a company suspected of supplying fake websites to criminal groups.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Catalin Cimpanu talks with Luke Wescott, Threat Detection Engineer at Sublime Security. Luke goes over the recent rise of calendar-based phishing campaigns over the past year and the reasons behind it.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI149.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI149/&quot;&gt;Sponsored: Sublime Security on the rise of ISC calendar phishing&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 12:03 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Sublime Security on the rise of ISC calendar phishing&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss CISO&apos;s expectations and priorities regarding budgets and what they will spend their money on. AI features heavily in 2027 priorities, but what will it deliver?</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN188.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN188/&quot;&gt;Between Two Nerds: Finding bugs does not mean security&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 31:10 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Finding bugs does not mean security&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/uV3Ek0ivrdo?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Finding bugs does not mean security"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:&#xa0;</h2><p><strong>Hackers lock out and extort Li Auto EV owners in Russia: </strong>Hackers are taking over the master user accounts of Li Auto smart EV cars sold in Russia, locking owners out of their cars, and extorting them to regain control of their vehicles.</p><p>The attacks have been <a href="https://www.autonews.ru/news/6ae87edd9a79478d050ecab3"><u>confirmed</u></a> to a Russian auto news portal by Alexander Rasfairy, the co-founder of a Li Auto car club. Rasfairy says at least 20 owners confirmed hacks, but the number of hijacked cars is likely much larger if considering non-members.</p><p>Most of the owners who had cars hijacked used Chinese eSIMs provided by the eSender service and by car importers to register the master car accounts.</p><p>[<a href="https://news.risky.biz/risky-bulletin-hackers-lock-out-and-extort-li-auto-smart-car-owners-in-russia/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Game devs are being extorted with AI game-cloning threats:</strong> For the past several months, mysterious parties have been sending emails to small game developers and threatening to copy their public demos and unreleased games using AI tools and publish them in advance unless they are paid a ransom demand.</p><p>Extortion fees have varied between hundreds and thousands of US dollars.</p><p>Some of the extortion emails have <a href="https://www.reddit.com/r/IndieDev/comments/1wtjcgf/someone_recreated_my_game_with_ai_and_is_now/"><u>made</u></a> <a href="https://www.reddit.com/r/IndieDev/comments/1wv9zbv/someone_is_saying_they_stole_my_game_and_is/"><u>their</u></a> <a href="https://x.com/highmotivation8/status/2111598637260427478"><u>way</u></a> <a href="https://x.com/Sini_Ming/status/2111297746640744522"><u>online</u></a> over the past weeks, but the campaign is believed to be the primary reason a large number of game developers <a href="https://www.youtube.com/watch?v=MRO1iaEv-LE"><u>started pulling demos</u></a> from the <a href="https://x.com/AllianceArtsEN/status/2111043223326163101"><u>Steam</u></a> <a href="https://x.com/DigitalSunGames/status/2112167974538723563"><u>platform</u></a>.</p><p>[<a href="https://news.risky.biz/risky-bulletin-game-devs-are-being-extorted-with-ai-game-cloning-threats/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Belarusian state hackers get on the data-wiping train: </strong>Hackers linked to the Belarusian government have conducted their first-ever destructive data-wiping attack in a major operational escalation.</p><p>According to <a href="https://www.welivesecurity.com/en/eset-research/eset-apt-activity-report-q2-2026-q3-2026/"><u>ESET&apos;s Q2-Q3 APT activity report</u></a>, the attack&#xa0; targeted a Ukrainian manufacturing company with an important role in the country&apos;s transportation sector.</p><p>ESET linked the attack to a group it tracks as FrostyNeighbor, but is more commonly known as GhostWriter. The security firm says Belarusian cyber groups have increasingly been helping Russia with its operations in Ukraine.</p><p>[<a href="https://news.risky.biz/risky-bulletin-belarusian-state-hackers-get-on-the-data-wiping-train/"><u>more</u></a> on<em> Risky Bulletin</em>]</p>
  109. Investigating GitHub Actions Access to an AWS Role

    Wed, 04 Nov 2026 00:00:00 -0000

    You are handed one record. It is an AssumeRoleWithWebIdentity event in a production AWS account, the role in it is named something like example-deploy-role, and the subject value starts with repo:. Somewhere in a GitHub organization you do not administer, a workflow job got an hour of credentials in your account. The ticket says “confirm scope.”
    <p>You are handed one record. It is an <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code> event in a production AWS account, the role in it is named something like <code class="language-plaintext highlighter-rouge">example-deploy-role</code>, and the subject value starts with <code class="language-plaintext highlighter-rouge">repo:</code>. Somewhere in a GitHub organization you do not administer, a workflow job got an hour of credentials in your account. The ticket says “confirm scope.”</p> <p>Three questions follow, and they are not the same question: which GitHub execution obtained this session, why did AWS admit it, and what did it actually do? Each one is answered by different evidence, held in a different system, with different retention. Working them in order is most of the job, and most of the work is knowing exactly what each field means and where it stops meaning anything.</p> <p>This walkthrough follows that path from the AWS side back to GitHub and forward again to resource activity. By the end you should be able to write a scope statement that separates what the records establish from what they merely suggest - the difference between a finding and a guess.</p> <p><em>Editorial Note: every account number, role ID, access key label, repository, run number and timestamp below is invented for this article. Account <code class="language-plaintext highlighter-rouge">111122223333</code> and bucket <code class="language-plaintext highlighter-rouge">amzn-s3-demo-bucket</code> are AWS documentation placeholders, not claimed to be reserved or real. No AWS account, GitHub organization or token was queried, minted or decoded in writing it. The synthetic records are trimmed to an allowlist of investigative fields: illustrations of field semantics, not observed evidence of anything.</em></p> <p><br /></p> <p><strong>Start from a successful issuance, not from the event name</strong></p> <p>An <code class="language-plaintext highlighter-rouge">eventName</code> of <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code> tells you a request arrived. It does not tell you a session existed. The operation <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">returns a set of temporary security credentials</a> consisting of an access key ID, a secret access key and a security token, so the thing you need in front of you is the <em>response</em>, not just the request.</p> <p>That is unusually available here. CloudTrail normally redacts <code class="language-plaintext highlighter-rouge">responseElements</code> for read-only operations, but the STS role-assumption operations are a documented exception: for <code class="language-plaintext highlighter-rouge">AssumeRole</code>, <code class="language-plaintext highlighter-rouge">AssumeRoleWithSAML</code> and <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code>, <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">CloudTrail includes the full <code class="language-plaintext highlighter-rouge">responseElements</code> except <code class="language-plaintext highlighter-rouge">secretAccessKey</code></a>. Read that twice before you export anything. The secret access key is excluded; the session token is not, and AWS’s own published example event shows a <code class="language-plaintext highlighter-rouge">sessionToken</code> placeholder in the response object. Treat the raw record as credential-bearing material, pull an allowlist of fields into your case notes, and leave the rest where it is.</p> <p>Here is the shape of what you keep.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"eventVersion"</span><span class="p">:</span><span class="w"> </span><span class="s2">"1.11"</span><span class="p">,</span><span class="w"> </span><span class="nl">"userIdentity"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"type"</span><span class="p">:</span><span class="w"> </span><span class="s2">"WebIdentityUser"</span><span class="p">,</span><span class="w"> </span><span class="nl">"principalId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com:sts.amazonaws.com:repo:example-org/example-repo:environment:prod"</span><span class="p">,</span><span class="w"> </span><span class="nl">"userName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"repo:example-org/example-repo:environment:prod"</span><span class="p">,</span><span class="w"> </span><span class="nl">"identityProvider"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"eventTime"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:47Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventSource"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts.amazonaws.com"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"AssumeRoleWithWebIdentity"</span><span class="p">,</span><span class="w"> </span><span class="nl">"awsRegion"</span><span class="p">:</span><span class="w"> </span><span class="s2">"us-west-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"requestParameters"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"roleArn"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:role/example-deploy-role"</span><span class="p">,</span><span class="w"> </span><span class="nl">"roleSessionName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"700001-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"durationSeconds"</span><span class="p">:</span><span class="w"> </span><span class="mi">3600</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"responseElements"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"credentials"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"accessKeyId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"SYNTHETIC_SESSION_KEY_01"</span><span class="p">,</span><span class="w"> </span><span class="nl">"expiration"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Apr 14, 2026, 10:12:47 AM"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"subjectFromWebIdentityToken"</span><span class="p">:</span><span class="w"> </span><span class="s2">"repo:example-org/example-repo:environment:prod"</span><span class="p">,</span><span class="w"> </span><span class="nl">"assumedRoleUser"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"assumedRoleId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"SYNTHETIC_ROLE_ID_01:700001-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"arn"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:sts::111122223333:assumed-role/example-deploy-role/700001-2"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"provider"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com"</span><span class="p">,</span><span class="w"> </span><span class="nl">"audience"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts.amazonaws.com"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"additionalEventData"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"identityProviderConnectionVerificationMethod"</span><span class="p">:</span><span class="w"> </span><span class="s2">"IAMTrustStore"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"eventID"</span><span class="p">:</span><span class="w"> </span><span class="s2">"00000000-0000-4000-8000-00000000f001"</span><span class="p">,</span><span class="w"> </span><span class="nl">"readOnly"</span><span class="p">:</span><span class="w"> </span><span class="kc">true</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventType"</span><span class="p">:</span><span class="w"> </span><span class="s2">"AwsApiCall"</span><span class="p">,</span><span class="w"> </span><span class="nl">"managementEvent"</span><span class="p">:</span><span class="w"> </span><span class="kc">true</span><span class="p">,</span><span class="w"> </span><span class="nl">"recipientAccountId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"111122223333"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventCategory"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Management"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p><em>Editorial Note: synthetic and deliberately incomplete. The <code class="language-plaintext highlighter-rouge">secretAccessKey</code> and <code class="language-plaintext highlighter-rouge">sessionToken</code> fields are omitted rather than redacted, and <code class="language-plaintext highlighter-rouge">SYNTHETIC_SESSION_KEY_01</code> and <code class="language-plaintext highlighter-rouge">SYNTHETIC_ROLE_ID_01</code> are nonfunctional labels, not key-shaped strings.</em></p> <p>Four fields carry most of the weight. <code class="language-plaintext highlighter-rouge">subjectFromWebIdentityToken</code> is, <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">for OpenID Connect ID tokens, the value the identity provider returned as the token’s <code class="language-plaintext highlighter-rouge">sub</code> claim</a> - that is your GitHub identity string. <code class="language-plaintext highlighter-rouge">audience</code> is <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">the intended audience, also known as client ID, of the web identity token</a>. <code class="language-plaintext highlighter-rouge">assumedRoleUser</code> gives you the session ARN and assumed role ID. <code class="language-plaintext highlighter-rouge">credentials.accessKeyId</code> is the key the session will sign with, and it is your join key for the next step.</p> <p>Two temporal traps. CloudTrail’s <code class="language-plaintext highlighter-rouge">eventTime</code> is <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-record-contents.html">the date and time the request was completed, in coordinated universal time</a> - a completion stamp, not a start stamp, and the acquisition time of the credentials only in the loose sense. Record it separately from the time you pulled the record. And <code class="language-plaintext highlighter-rouge">eventID</code> is <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-record-contents.html">a GUID generated by CloudTrail to uniquely identify each event</a>. It identifies an event. It is not a session identifier, and there is no arithmetic that turns it into a GitHub run ID.</p> <p>Note <code class="language-plaintext highlighter-rouge">provider</code> here. The STS API reference defines the response <code class="language-plaintext highlighter-rouge">Provider</code> as <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">the issuing authority of the web identity token, containing the value of the <code class="language-plaintext highlighter-rouge">iss</code> field for OpenID Connect ID tokens</a> - which would be the issuer URL. AWS’s own OIDC CloudTrail example instead shows <code class="language-plaintext highlighter-rouge">provider</code> holding an IAM OIDC-provider ARN, and <code class="language-plaintext highlighter-rouge">userIdentity.identityProvider</code> likewise, while the CloudTrail identity reference says that for a <code class="language-plaintext highlighter-rouge">WebIdentityUser</code> the <code class="language-plaintext highlighter-rouge">identityProvider</code> field <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-user-identity.html">is the issuer name of the web identity federation provider</a>. Those two representations do not agree, and I have not found a document that reconciles them. So do not hard-code an expected format: copy what your record contains, verbatim, and corroborate it against the account’s IAM OIDC provider objects. The <code class="language-plaintext highlighter-rouge">identityProviderConnectionVerificationMethod</code> value under <code class="language-plaintext highlighter-rouge">additionalEventData</code> is a useful companion - it records <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">whether AWS verified the connection using its library of trusted root CAs (<code class="language-plaintext highlighter-rouge">IAMTrustStore</code>) or a configured certificate thumbprint (<code class="language-plaintext highlighter-rouge">Thumbprint</code>)</a>.</p> <p>Finally, <code class="language-plaintext highlighter-rouge">readOnly</code> is <code class="language-plaintext highlighter-rouge">true</code> on this event, and it is worth saying what that does not mean. It classifies <em>this control-plane operation</em> as non-mutating. The credentials it handed out are a different thing, and every request made with them gets evaluated on its own against whatever policies applied at that moment - a later section’s problem. Do not filter STS issuance out of an investigation because a read-only flag made it look boring.</p> <p><br /></p> <p><strong>Join the session to downstream requests, and know when you cannot</strong></p> <p>The strongest available link between issuance and use is the access key. In a CloudTrail <code class="language-plaintext highlighter-rouge">userIdentity</code>, <code class="language-plaintext highlighter-rouge">accessKeyId</code> is <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-user-identity.html">the access key ID that was used to sign the request, and for temporary credentials it is the access key ID of the temporary credentials</a>. So <code class="language-plaintext highlighter-rouge">responseElements.credentials.accessKeyId</code> from the issuance event should appear as <code class="language-plaintext highlighter-rouge">userIdentity.accessKeyId</code> on requests made with that session.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"eventVersion"</span><span class="p">:</span><span class="w"> </span><span class="s2">"1.11"</span><span class="p">,</span><span class="w"> </span><span class="nl">"userIdentity"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"type"</span><span class="p">:</span><span class="w"> </span><span class="s2">"AssumedRole"</span><span class="p">,</span><span class="w"> </span><span class="nl">"principalId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"SYNTHETIC_ROLE_ID_01:700001-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"arn"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:sts::111122223333:assumed-role/example-deploy-role/700001-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"accountId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"111122223333"</span><span class="p">,</span><span class="w"> </span><span class="nl">"accessKeyId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"SYNTHETIC_SESSION_KEY_01"</span><span class="p">,</span><span class="w"> </span><span class="nl">"sessionContext"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"sessionIssuer"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"type"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Role"</span><span class="p">,</span><span class="w"> </span><span class="nl">"principalId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"SYNTHETIC_ROLE_ID_01"</span><span class="p">,</span><span class="w"> </span><span class="nl">"arn"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:role/example-deploy-role"</span><span class="p">,</span><span class="w"> </span><span class="nl">"accountId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"111122223333"</span><span class="p">,</span><span class="w"> </span><span class="nl">"userName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"example-deploy-role"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"webIdFederationData"</span><span class="p">:</span><span class="w"> </span><span class="p">{},</span><span class="w"> </span><span class="nl">"attributes"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"creationDate"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:47Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"mfaAuthenticated"</span><span class="p">:</span><span class="w"> </span><span class="s2">"false"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"eventTime"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:13:22Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventSource"</span><span class="p">:</span><span class="w"> </span><span class="s2">"s3.amazonaws.com"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"PutObject"</span><span class="p">,</span><span class="w"> </span><span class="nl">"awsRegion"</span><span class="p">:</span><span class="w"> </span><span class="s2">"us-west-2"</span><span class="p">,</span><span class="w"> </span><span class="nl">"requestParameters"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"bucketName"</span><span class="p">:</span><span class="w"> </span><span class="s2">"amzn-s3-demo-bucket"</span><span class="p">,</span><span class="w"> </span><span class="nl">"key"</span><span class="p">:</span><span class="w"> </span><span class="s2">"deploy/example.txt"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"responseElements"</span><span class="p">:</span><span class="w"> </span><span class="kc">null</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventID"</span><span class="p">:</span><span class="w"> </span><span class="s2">"00000000-0000-4000-8000-00000000f002"</span><span class="p">,</span><span class="w"> </span><span class="nl">"readOnly"</span><span class="p">:</span><span class="w"> </span><span class="kc">false</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventType"</span><span class="p">:</span><span class="w"> </span><span class="s2">"AwsApiCall"</span><span class="p">,</span><span class="w"> </span><span class="nl">"managementEvent"</span><span class="p">:</span><span class="w"> </span><span class="kc">false</span><span class="p">,</span><span class="w"> </span><span class="nl">"recipientAccountId"</span><span class="p">:</span><span class="w"> </span><span class="s2">"111122223333"</span><span class="p">,</span><span class="w"> </span><span class="nl">"eventCategory"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Data"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p>Three identifier checks sit alongside the key match, then one timing check that needs care. The session ARN matches <code class="language-plaintext highlighter-rouge">responseElements.assumedRoleUser.arn</code>. The <code class="language-plaintext highlighter-rouge">principalId</code> is the assumed role ID plus session name, matching <code class="language-plaintext highlighter-rouge">assumedRoleId</code>. <code class="language-plaintext highlighter-rouge">sessionContext.sessionIssuer</code> names the issuing role, with its own stable <code class="language-plaintext highlighter-rouge">principalId</code>, <code class="language-plaintext highlighter-rouge">arn</code> and <code class="language-plaintext highlighter-rouge">accountId</code>.</p> <p>Then <code class="language-plaintext highlighter-rouge">sessionContext.attributes.creationDate</code>, which is <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-user-identity.html">the date and time when the temporary security credentials were issued</a>. Do not require it to equal the issuance record’s <code class="language-plaintext highlighter-rouge">eventTime</code>: they are not the same quantity, the other being <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-record-contents.html">the date and time the request was completed</a>. What you want is compatibility at the precision your records actually carry, with a <code class="language-plaintext highlighter-rouge">creationDate</code> at or just before the issuance <code class="language-plaintext highlighter-rouge">eventTime</code>. In the two synthetic records here they are byte-identical because I set them that way - a convenience of the example, not a matching rule to code against. A few seconds of daylight is not a reason to reject a join the access key already supports.</p> <p>And none of these four are independent witnesses. They are fields AWS derived from one role assumption and carried into two records, so they corroborate the key match rather than replacing it.</p> <p>Now the counterexamples, because this is where investigations quietly go wrong.</p> <ul> <li><strong>The issuance <code class="language-plaintext highlighter-rouge">principalId</code> is not the downstream <code class="language-plaintext highlighter-rouge">principalId</code>.</strong> On the STS event the identity type is <code class="language-plaintext highlighter-rouge">WebIdentityUser</code>, whose <code class="language-plaintext highlighter-rouge">principalId</code> is documented as <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-user-identity.html">a combination of the issuer, application ID, and user ID</a>. On the S3 event the type is <code class="language-plaintext highlighter-rouge">AssumedRole</code> and the <code class="language-plaintext highlighter-rouge">principalId</code> is the assumed role ID plus session name. Those are different namespaces; joining one to the other returns nothing.</li> <li><strong>The access key can be absent.</strong> For security reasons <code class="language-plaintext highlighter-rouge">accessKeyId</code> <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-user-identity.html">might not be present, or might be displayed as an empty string</a>. When that happens you fall back to session ARN plus time window, which is weaker - say so rather than quietly promoting it to the same confidence.</li> <li><strong>The session name is caller-controlled and reusable.</strong> <code class="language-plaintext highlighter-rouge">RoleSessionName</code> is simply <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">an identifier for the assumed role session</a> supplied by the requester, and it is what gets embedded in the ARN and assumed role ID. If two jobs in one run use the same name, the session ARNs are identical and only the access key and timestamps separate them.</li> <li><strong>A credential match is not a person.</strong> For <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code>, CloudTrail records the role’s subsequent API calls with <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">role identity only, no user</a>. You have established which credential set signed the request; who or what drove it is a separate question, answered on the GitHub side and imperfectly.</li> </ul> <p>Bound the activity window with the issuance time and the expiration from the response, then look for anything signed by that key inside it. Session duration matters here: the default is <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">3600 seconds, and the ceiling is the role’s own maximum session duration setting, which can be one to twelve hours</a>.</p> <p><br /></p> <p><strong>Identify the GitHub run, attempt and job</strong></p> <p>The subject in the record - <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:environment:prod</code> - tells you the repository and the job context. It does not tell you which run. For that you need GitHub-side evidence, and the fastest lead is the session name, if whoever built the workflow was thoughtful enough to make it mean something. In this case it is <code class="language-plaintext highlighter-rouge">700001-2</code>, which the workflow set from the run ID and attempt number. That is a lead, not proof; the value is caller-supplied.</p> <p>Take the lead to the documented read routes. <code class="language-plaintext highlighter-rouge">GET /repos/{owner}/{repo}/actions/runs/{run_id}</code> <a href="https://docs.github.com/en/rest/actions/workflow-runs">gets a specific workflow run</a>, <code class="language-plaintext highlighter-rouge">GET /repos/{owner}/{repo}/actions/runs/{run_id}/attempts/{attempt_number}</code> <a href="https://docs.github.com/en/rest/actions/workflow-runs">gets a specific workflow run attempt</a>, and <code class="language-plaintext highlighter-rouge">GET /repos/{owner}/{repo}/actions/runs/{run_id}/attempts/{attempt_number}/jobs</code> <a href="https://docs.github.com/en/rest/actions/workflow-jobs">lists jobs for that attempt</a>. GitHub documents all three as usable by <a href="https://docs.github.com/en/rest/actions/workflow-runs">anyone with read access to the repository</a> - which is true, and is not the whole requirement. Against a private repository the same pages add that <a href="https://docs.github.com/en/rest/actions/workflow-jobs">OAuth app tokens and personal access tokens (classic) need the <code class="language-plaintext highlighter-rouge">repo</code> scope</a>. That sentence covers classic tokens; I haven’t verified the fine-grained-token requirements here. Confirm the requirements for your token type with whoever administers the organization before a case is on the clock.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"run"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"id"</span><span class="p">:</span><span class="w"> </span><span class="mi">700001</span><span class="p">,</span><span class="w"> </span><span class="nl">"run_attempt"</span><span class="p">:</span><span class="w"> </span><span class="mi">2</span><span class="p">,</span><span class="w"> </span><span class="nl">"run_number"</span><span class="p">:</span><span class="w"> </span><span class="mi">184</span><span class="p">,</span><span class="w"> </span><span class="nl">"workflow_id"</span><span class="p">:</span><span class="w"> </span><span class="mi">4410077</span><span class="p">,</span><span class="w"> </span><span class="nl">"path"</span><span class="p">:</span><span class="w"> </span><span class="s2">".github/workflows/deploy.yml"</span><span class="p">,</span><span class="w"> </span><span class="nl">"event"</span><span class="p">:</span><span class="w"> </span><span class="s2">"push"</span><span class="p">,</span><span class="w"> </span><span class="nl">"head_branch"</span><span class="p">:</span><span class="w"> </span><span class="s2">"main"</span><span class="p">,</span><span class="w"> </span><span class="nl">"head_sha"</span><span class="p">:</span><span class="w"> </span><span class="s2">"9f1c2ad4e6b8c0d2f4a6b8c0d2e4f6a8b0c2d4e6"</span><span class="p">,</span><span class="w"> </span><span class="nl">"run_started_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:11:40Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"previous_attempt_url"</span><span class="p">:</span><span class="w"> </span><span class="s2">"https://api.github.com/repos/example-org/example-repo/actions/runs/700001/attempts/1"</span><span class="p">,</span><span class="w"> </span><span class="nl">"actor"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"login"</span><span class="p">:</span><span class="w"> </span><span class="s2">"example-author"</span><span class="p">,</span><span class="w"> </span><span class="nl">"id"</span><span class="p">:</span><span class="w"> </span><span class="mi">500101</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"triggering_actor"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"login"</span><span class="p">:</span><span class="w"> </span><span class="s2">"example-releaser"</span><span class="p">,</span><span class="w"> </span><span class="nl">"id"</span><span class="p">:</span><span class="w"> </span><span class="mi">500102</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"referenced_workflows"</span><span class="p">:</span><span class="w"> </span><span class="p">[],</span><span class="w"> </span><span class="nl">"repository"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"id"</span><span class="p">:</span><span class="w"> </span><span class="mi">600200</span><span class="p">,</span><span class="w"> </span><span class="nl">"full_name"</span><span class="p">:</span><span class="w"> </span><span class="s2">"example-org/example-repo"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"job"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"id"</span><span class="p">:</span><span class="w"> </span><span class="mi">9900012345</span><span class="p">,</span><span class="w"> </span><span class="nl">"run_id"</span><span class="p">:</span><span class="w"> </span><span class="mi">700001</span><span class="p">,</span><span class="w"> </span><span class="nl">"run_attempt"</span><span class="p">:</span><span class="w"> </span><span class="mi">2</span><span class="p">,</span><span class="w"> </span><span class="nl">"name"</span><span class="p">:</span><span class="w"> </span><span class="s2">"publish"</span><span class="p">,</span><span class="w"> </span><span class="nl">"status"</span><span class="p">:</span><span class="w"> </span><span class="s2">"completed"</span><span class="p">,</span><span class="w"> </span><span class="nl">"conclusion"</span><span class="p">:</span><span class="w"> </span><span class="s2">"success"</span><span class="p">,</span><span class="w"> </span><span class="nl">"started_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:03Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"completed_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:13:51Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"runner_name"</span><span class="p">:</span><span class="w"> </span><span class="s2">"GitHub Actions 42"</span><span class="p">,</span><span class="w"> </span><span class="nl">"head_branch"</span><span class="p">:</span><span class="w"> </span><span class="s2">"main"</span><span class="p">,</span><span class="w"> </span><span class="nl">"steps"</span><span class="p">:</span><span class="w"> </span><span class="p">[</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"number"</span><span class="p">:</span><span class="w"> </span><span class="mi">2</span><span class="p">,</span><span class="w"> </span><span class="nl">"name"</span><span class="p">:</span><span class="w"> </span><span class="s2">"configure aws credentials"</span><span class="p">,</span><span class="w"> </span><span class="nl">"started_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:44Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"completed_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:49Z"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"number"</span><span class="p">:</span><span class="w"> </span><span class="mi">3</span><span class="p">,</span><span class="w"> </span><span class="nl">"name"</span><span class="p">:</span><span class="w"> </span><span class="s2">"upload build output"</span><span class="p">,</span><span class="w"> </span><span class="nl">"started_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:12:49Z"</span><span class="p">,</span><span class="w"> </span><span class="nl">"completed_at"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2026-04-14T09:13:40Z"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">]</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p><em>Editorial Note: a synthetic composite of fields from two separate documented responses, trimmed to what the correlation needs. It is not a verbatim API response, and GitHub did not log any of it into CloudTrail - you gather it yourself and record that you did.</em></p> <p>The correlation you are building is a timeline overlap, not a cryptographic binding. Step 2 ran from <code class="language-plaintext highlighter-rouge">09:12:44</code> to <code class="language-plaintext highlighter-rouge">09:12:49</code>; the STS issuance completed at <code class="language-plaintext highlighter-rouge">09:12:47</code>; the S3 write completed at <code class="language-plaintext highlighter-rouge">09:13:22</code>, inside step 3. Two AWS activity stamps, each landing inside the step that would have produced it - that is what you hoped for and worth writing down. The expiration at <code class="language-plaintext highlighter-rouge">10:12:47</code> is not a third one and does not fall inside the job at all; it is a lifetime bound that outlives the job by about an hour, which deserves its own line, because a credential that outlives the execution stays usable after it. Be careful how you describe the two that do fit. These are two systems each recording their own timestamps at their own precision, and unless you have established what time source both were using, containment is what you have - not agreement between synchronized clocks, and not independent confirmations of one fact. It stays circumstantial, and if a second job in the same window assumed the same role, you have two candidates and should report two.</p> <p><br /></p> <p><img width="1448" height="1086" alt="Forensic Cloud Credential Timeline" src="https://github.com/user-attachments/assets/7ef17c47-76f5-4e0d-b7b0-bbe2b0b3a3be" /></p> <p><br /></p> <p>Some identity semantics worth getting right before you write any of this down:</p> <ul> <li><strong>Re-runs keep the run ID.</strong> <code class="language-plaintext highlighter-rouge">github.run_id</code> is <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">a unique number for each workflow run within a repository, and this number does not change if you re-run the workflow run</a>. <code class="language-plaintext highlighter-rouge">github.run_attempt</code> <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">begins at 1 for the first attempt and increments with each re-run</a>. A run ID alone is therefore ambiguous across attempts, which is why the attempt-scoped job route exists.</li> <li><strong>A run and attempt still are not one execution.</strong> The pair does not identify a single job, a single matrix leg, or a single STS call: a workflow can assume the same role from several jobs, and a job can assume it more than once. So enumerate the attempt’s job list completely before ruling any job out. It is paginated - <a href="https://docs.github.com/en/rest/actions/workflow-jobs">30 results per page by default, 100 maximum</a> - and the response carries a <code class="language-plaintext highlighter-rouge">total_count</code> to reconcile against the rows you collected. Excluding a competing candidate on the strength of page one is how the second candidate disappears from the report.</li> <li><strong>The numeric job ID is not <code class="language-plaintext highlighter-rouge">github.job</code>.</strong> The REST <code class="language-plaintext highlighter-rouge">job.id</code> is an integer; <code class="language-plaintext highlighter-rouge">github.job</code> is the workflow’s job key, <code class="language-plaintext highlighter-rouge">publish</code> in this example. Keep both, and label them.</li> <li><strong>Actor and triggering actor can differ, and the privilege story is counterintuitive.</strong> GitHub documents that <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">re-runs use the privileges of <code class="language-plaintext highlighter-rouge">github.actor</code>, even if the actor initiating the re-run (<code class="language-plaintext highlighter-rouge">github.triggering_actor</code>) has different privileges</a>. That is GitHub execution context. It says nothing about AWS effective permissions, and neither login proves a human sat at a keyboard. Keep the numeric account IDs next to the logins, because logins are renameable.</li> </ul> <p>One more thing about that action’s retry behavior, because it produces a specific false lead. The pinned README for the AWS credentials action states that on a failed STS call it retries <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">with exponential backoff twelve times by default</a>, configurable through <code class="language-plaintext highlighter-rouge">retry-max-attempts</code> or disabled with <code class="language-plaintext highlighter-rouge">disable-retry</code>. So a cluster of STS records seconds apart has an innocent explanation available to it. Available is not the same as established: that explanation only holds if that action, at a revision carrying that behavior, is what actually ran, and it competes with a workflow that assumes the role repeatedly across jobs, steps or matrix legs.</p> <p>What it does not compete with is credential reuse. Each <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code> call requires another token presentation: the operation’s <code class="language-plaintext highlighter-rouge">WebIdentityToken</code> parameter - <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">the OAuth 2.0 access token or OpenID Connect ID token that is provided by the identity provider</a> - is required. Someone using credentials already issued produces downstream calls signed by that access key, not a second assumption. The record type narrows what happened; it still does not tell you who presented the token. Establish the executed revision and check <code class="language-plaintext highlighter-rouge">run_attempt</code> before you settle it either way. A retry story is a hypothesis you test, not a reason to stop counting.</p> <p><br /></p> <p><strong>Establish what executed, not what the branch says today</strong></p> <p>The repository’s <code class="language-plaintext highlighter-rouge">main</code> branch as it stands right now is not evidence about April. Four different things are commonly confused here, and they answer four different questions.</p> <div class="language-yaml highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="na">name</span><span class="pi">:</span> <span class="s">deploy</span> <span class="na">on</span><span class="pi">:</span> <span class="na">push</span><span class="pi">:</span> <span class="na">branches</span><span class="pi">:</span> <span class="pi">[</span><span class="nv">main</span><span class="pi">]</span> <span class="na">permissions</span><span class="pi">:</span> <span class="na">id-token</span><span class="pi">:</span> <span class="s">write</span> <span class="na">contents</span><span class="pi">:</span> <span class="s">read</span> <span class="na">jobs</span><span class="pi">:</span> <span class="na">publish</span><span class="pi">:</span> <span class="na">runs-on</span><span class="pi">:</span> <span class="s">ubuntu-latest</span> <span class="na">environment</span><span class="pi">:</span> <span class="s">prod</span> <span class="na">steps</span><span class="pi">:</span> <span class="pi">-</span> <span class="na">uses</span><span class="pi">:</span> <span class="s">actions/checkout@v6</span> <span class="pi">-</span> <span class="na">name</span><span class="pi">:</span> <span class="s">configure aws credentials</span> <span class="na">uses</span><span class="pi">:</span> <span class="s">aws-actions/configure-aws-credentials@e3dd6a429d7300a6a4c196c26e071d42e0343502</span> <span class="na">with</span><span class="pi">:</span> <span class="na">role-to-assume</span><span class="pi">:</span> <span class="s">arn:aws:iam::111122223333:role/example-deploy-role</span> <span class="na">role-session-name</span><span class="pi">:</span> <span class="s">$-$</span> <span class="na">aws-region</span><span class="pi">:</span> <span class="s">us-west-2</span> <span class="pi">-</span> <span class="na">name</span><span class="pi">:</span> <span class="s">upload build output</span> <span class="na">run</span><span class="pi">:</span> <span class="s">aws s3 cp ./build/example.txt s3://amzn-s3-demo-bucket/deploy/example.txt</span> </code></pre></div></div> <p><em>Editorial Note: an illustrative fragment showing the fields this investigation reads. It is not a deployment template and has not been executed. The pinned action commit is the one GitHub’s own AWS guide uses in its example workflow; inspecting that README is not an audit of the action’s compiled code.</em></p> <p>The workflow definition’s own revision is <code class="language-plaintext highlighter-rouge">github.workflow_sha</code>, <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">the commit SHA for the workflow file</a>. The triggering commit is <code class="language-plaintext highlighter-rouge">github.sha</code>, and GitHub is explicit that <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">the value of this commit SHA depends on the event that triggered the workflow</a> - for an unmerged pull request the documented <code class="language-plaintext highlighter-rouge">ref</code> is a merge ref, and <code class="language-plaintext highlighter-rouge">pull_request_target</code> events carry <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">the <code class="language-plaintext highlighter-rouge">ref</code> from the base branch</a>. What the job actually checked out is a third thing, decided by the checkout step. And the action revisions and any reusable workflows are a fourth; the run response carries <code class="language-plaintext highlighter-rouge">referenced_workflows</code> entries with their own <code class="language-plaintext highlighter-rouge">path</code>, <code class="language-plaintext highlighter-rouge">sha</code> and <code class="language-plaintext highlighter-rouge">ref</code> - exactly what you want when a caller workflow is doing the real work.</p> <p>Be careful about promising yourself a field you may never have. <code class="language-plaintext highlighter-rouge">workflow_sha</code> is a documented context value and an <a href="https://docs.github.com/en/actions/reference/security/oidc">OIDC token claim</a>; it is not a documented member of the REST workflow run response schema, which carries <code class="language-plaintext highlighter-rouge">path</code>, <code class="language-plaintext highlighter-rouge">workflow_id</code> and <code class="language-plaintext highlighter-rouge">head_sha</code> instead. Get the definition’s revision from retained run evidence rather than assuming an API hands it over.</p> <p>Tags are moving pointers. <code class="language-plaintext highlighter-rouge">actions/checkout@v6</code> is a tag, and resolving that tag today tells you what it points at today. The revision that ran in April has to come from evidence retained from April, and the attempt logs are the usual place to look: GitHub documents a <a href="https://docs.github.com/en/rest/actions/workflow-runs">redirect URL to download an archive of log files for a specific workflow run attempt</a>, and that link expires after one minute, so follow the redirect programmatically rather than pasting it into a browser tab you will get back to in a moment.</p> <p>Two things that endpoint does not promise. It does not promise the logs are still there - retention is a separate setting, and <a href="https://docs.github.com/en/rest/actions/workflow-runs">deleting all logs for a workflow run</a> is its own documented route, so absence of logs is not evidence about what ran. And it does not promise that every executed revision was ever written into them; a step that resolves an action to a commit without printing it leaves you nothing to read. When it is unrecoverable, record it as unrecoverable from retained evidence. Do not solve the gap with a fresh run or a fresh OIDC token; a new token is evidence about now.</p> <p>And resist the urge to dump the whole <code class="language-plaintext highlighter-rouge">github</code> context into the job log to fill in blanks. GitHub warns that <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">certain contexts should be treated as untrusted input</a>, and <code class="language-plaintext highlighter-rouge">github.token</code> is <a href="https://docs.github.com/en/actions/reference/workflows-and-actions/contexts">a token to authenticate on behalf of the GitHub App installed on your repository</a> - a third credential, distinct from the OIDC JWT and the AWS session. Three credentials, three blast radii, and only one is what you are investigating.</p> <p>While we are on that: <code class="language-plaintext highlighter-rouge">id-token: write</code> is the permission that lets the job get the JWT at all, and GitHub is blunt that <a href="https://docs.github.com/en/actions/reference/security/oidc">this setting only enables fetching and setting the OIDC token; it does not grant write access to other resources</a>. Seeing it in a workflow file is not a finding.</p> <p><br /></p> <p><img width="863" height="569" alt="GitHub AWS Evidence Correlation Flowchart" src="https://github.com/user-attachments/assets/f34d9679-69af-44fa-9daa-841403c443f5" /></p> <p><br /></p> <p><strong>Work out why AWS admitted the subject - at the time, not today</strong></p> <p>Now the trust layer. Three things had to be true: an OIDC provider had to exist and validate the token, the role’s trust policy had to allow this principal and action, and its conditions had to match the presented claims.</p> <p>The provider is account-local. AWS notes that <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_providers_create_oidc.html">OIDC IdPs used in a role trust policy must be in the same account as the role that trusts it</a>, and the provider registration carries the issuer URL and the audience list. GitHub’s AWS guide says to register the <code class="language-plaintext highlighter-rouge">token.actions.githubusercontent.com</code> issuer as the provider URL and to use <a href="https://docs.github.com/en/actions/how-tos/secure-your-work/security-harden-deployments/oidc-in-aws"><code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code> as the audience if you are using the official action</a>. That is not GitHub’s own default: GitHub’s OIDC reference says the <code class="language-plaintext highlighter-rouge">aud</code> claim <a href="https://docs.github.com/en/actions/reference/security/oidc">is by default the URL of the repository owner</a>, and the action’s README states that <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">by default the audience is <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code></a>, with a documented <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md"><code class="language-plaintext highlighter-rouge">sts.amazonaws.com.cn</code> example for a China-region audience</a>. So an <code class="language-plaintext highlighter-rouge">audience</code> of <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code> in your record is the AWS action’s convention in the commercial partition, and any other value - an owner URL, a different partition’s endpoint - is worth a second look.</p> <p>Then the trust policy, which is not the same document as the role’s permissions. AWS separates them explicitly: one is <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_create_for-idp_oidc.html">the trust policy that specifies who can assume the role</a>, the other is <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_create_for-idp_oidc.html">the permissions policy that specifies the AWS actions and resources</a> the assumed identity may use. Conflating them produces confident wrong answers in both directions.</p> <p>Here is the trust document this synthetic case stipulates was in force from 2026-02-03. Note what that sentence is doing: in a real investigation a dated historical policy is an artifact you have to go and find, and stipulating one is the move you are not allowed to make.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"Version"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2012-10-17"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Statement"</span><span class="p">:</span><span class="w"> </span><span class="p">[</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"Sid"</span><span class="p">:</span><span class="w"> </span><span class="s2">"GitHubDeploy"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Effect"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Allow"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Principal"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"Federated"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"Action"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts:AssumeRoleWithWebIdentity"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Condition"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"StringEquals"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"token.actions.githubusercontent.com:aud"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts.amazonaws.com"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"StringLike"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"token.actions.githubusercontent.com:sub"</span><span class="p">:</span><span class="w"> </span><span class="s2">"repo:example-org/example-repo:*"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">]</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p>And here is the <code class="language-plaintext highlighter-rouge">Role.AssumeRolePolicyDocument</code> extracted from a <code class="language-plaintext highlighter-rouge">GetRole</code> capture taken on 2026-09-19, five months after the session. One field out of that response, not the response - and decoded conditionally rather than reflexively. AWS notes that <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role.html">policies returned by this operation are URL-encoded compliant with RFC 3986</a> and, in the same note, that <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role.html">some SDKs do this decoding automatically</a>. So look at what you actually got back and decode only if it is still encoded; running a decode over an already-decoded document is how a <code class="language-plaintext highlighter-rouge">+</code> inside a value quietly becomes a space.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"Version"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2012-10-17"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Statement"</span><span class="p">:</span><span class="w"> </span><span class="p">[</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"Sid"</span><span class="p">:</span><span class="w"> </span><span class="s2">"GitHubDeploy"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Effect"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Allow"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Principal"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"Federated"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:iam::111122223333:oidc-provider/token.actions.githubusercontent.com"</span><span class="w"> </span><span class="p">},</span><span class="w"> </span><span class="nl">"Action"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts:AssumeRoleWithWebIdentity"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Condition"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"StringEquals"</span><span class="p">:</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"token.actions.githubusercontent.com:aud"</span><span class="p">:</span><span class="w"> </span><span class="s2">"sts.amazonaws.com"</span><span class="p">,</span><span class="w"> </span><span class="nl">"token.actions.githubusercontent.com:sub"</span><span class="p">:</span><span class="w"> </span><span class="s2">"repo:example-org/example-repo:ref:refs/heads/main"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">]</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p>If you only had the second document you would reach a contradiction: the job ran on <code class="language-plaintext highlighter-rouge">main</code>, the current policy allows exactly <code class="language-plaintext highlighter-rouge">main</code>, and yet the recorded subject is an environment subject that this policy does not match. The trap is temporal. <code class="language-plaintext highlighter-rouge">GetRole</code> <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role.html">retrieves information about the specified role</a> as it stands when you ask, and the <code class="language-plaintext highlighter-rouge">CreateDate</code> it returns is <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role.html">when the role was created</a> - not when the trust policy last changed. The change itself is a separate, findable event: <code class="language-plaintext highlighter-rouge">UpdateAssumeRolePolicy</code> <a href="https://docs.aws.amazon.com/IAM/latest/APIReference/API_UpdateAssumeRolePolicy.html">updates the policy that grants an IAM entity permission to assume a role</a>.</p> <p>Finding those events is necessary. It is not sufficient. Reconstructing the policy <em>this way</em> - a dated baseline carried forward through the changes recorded after it - needs four things, and a gap in any one of them leaves this particular reconstruction incomplete rather than merely approximate: a dated baseline document you actually hold, the <em>successful</em> change events between that baseline and the issuance, a coverage statement establishing that your search would have surfaced those events had they existed, and the identity of the role itself. That is a requirement list for this route, not a claim that it is the only credible one; the alternatives below have their own, different requirements. On that last point, <code class="language-plaintext highlighter-rouge">GetRole</code> returns <code class="language-plaintext highlighter-rouge">RoleId</code>, <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role.html">the stable and unique string identifying the role</a>, which is what tells you whether a role deleted and recreated under the same ARN is the same role. Every ARN in your notes looks identical either way.</p> <p>Two things you must not do with this. First, do not describe the current narrow policy as the explanation for an earlier environment-subject issuance. It is a later state of the same role, and what the pair establishes is that the scope changed somewhere between those two dates. Calling the tightening a remediation asserts a cause and a motive you have not evidenced - an unrelated cleanup, a template rollout or a different incident entirely all produce the same pair of documents. Second, do not describe an absent change event as proof nothing changed. That depends on collection coverage, which is the last section of this article.</p> <p>One more place people stop too early. IAM not versioning a document is a statement about IAM, not about the world. A trust policy or an inline policy may still be recoverable from a configuration-management snapshot, a change-management record, or the repository that defines the infrastructure - and if you go there, keep the distinction between a committed definition and the deployed state. A merged commit is a proposed change with a date on it; whether it was ever applied to that role, and when, the commit cannot answer.</p> <p><br /></p> <p><strong>The subject format question, and a matrix that settles it</strong></p> <p>GitHub builds the default subject by concatenating workflow metadata, and the precedence rules are the part people get wrong. The subject <a href="https://docs.github.com/en/actions/reference/security/oidc">includes the environment name when the job references an environment</a>. It <a href="https://docs.github.com/en/actions/reference/security/oidc">includes the branch name only if the job doesn’t reference an environment, and if the workflow is not triggered by a pull request event</a>. And it includes <a href="https://docs.github.com/en/actions/reference/security/oidc">the <code class="language-plaintext highlighter-rouge">pull_request</code> string when the workflow is triggered by a pull request event, but only if the job doesn’t reference an environment</a>.</p> <p>Those are mutually exclusive suffixes. A job on <code class="language-plaintext highlighter-rouge">main</code> that declares <code class="language-plaintext highlighter-rouge">environment: prod</code> produces <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:environment:prod</code>. It does not produce a subject containing both the branch and the environment, and GitHub’s AWS guide agrees from the other direction: if you use a workflow with an environment, <a href="https://docs.github.com/en/actions/how-tos/secure-your-work/security-harden-deployments/oidc-in-aws">the <code class="language-plaintext highlighter-rouge">sub</code> field must reference the environment name</a>. That rule is why a <code class="language-plaintext highlighter-rouge">main</code>-branch job here does not match a <code class="language-plaintext highlighter-rouge">main</code>-branch condition, and it is the first thing to check whenever a recorded subject and a written condition look like they ought to agree and do not.</p> <p>The matching itself is ordinary string comparison, not a regex engine. <code class="language-plaintext highlighter-rouge">StringEquals</code> is <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_elements_condition_operators.html#Conditions_String">exact matching, case sensitive</a>. <code class="language-plaintext highlighter-rouge">StringLike</code> is <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_elements_condition_operators.html#Conditions_String">case-sensitive matching whose values can include multi-character match wildcards (<code class="language-plaintext highlighter-rouge">*</code>) and single-character match wildcards (<code class="language-plaintext highlighter-rouge">?</code>) anywhere in the string</a>. And absence is not neutrality: <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_elements_condition_operators.html">if the key that you specify in a policy condition is not present in the request context, the values do not match and the condition is false</a>.</p> <p><br /></p> <p><img width="877" height="389" alt="GitHub OIDC Subject Matching" src="https://github.com/user-attachments/assets/04c93d5b-e563-442c-9661-b4995b900d05" /></p> <p><br /></p> <p>Apply both <code class="language-plaintext highlighter-rouge">Condition</code> blocks above to the same claim pairs:</p> <table> <thead> <tr> <th>Claims presented</th> <th>Narrow (<code class="language-plaintext highlighter-rouge">StringEquals</code> on main)</th> <th>Broad (<code class="language-plaintext highlighter-rouge">StringLike</code> on repo)</th> </tr> </thead> <tbody> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:ref:refs/heads/main</code></td> <td>match</td> <td>match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:ref:refs/heads/feature</code></td> <td>no match</td> <td>match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:pull_request</code></td> <td>no match</td> <td>match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:environment:prod</code></td> <td>no match</td> <td>match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/other-repo:ref:refs/heads/main</code></td> <td>no match</td> <td>no match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com.cn</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:ref:refs/heads/main</code></td> <td>no match</td> <td>no match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> <code class="language-plaintext highlighter-rouge">repo:Example-Org/example-repo:ref:refs/heads/main</code></td> <td>no match</td> <td>no match</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">aud</code> <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, <code class="language-plaintext highlighter-rouge">sub</code> key absent from request context</td> <td>no match</td> <td>no match</td> </tr> </tbody> </table> <p><em>Editorial Note: every row uses the legacy default subject format, which this case stipulates for the April repository - no subject customization, no immutable-format transition, and an issuance date preceding the July 15, 2026 transition date GitHub documents. I computed these results offline with ordinary case-sensitive string and prefix-glob matching against the two <code class="language-plaintext highlighter-rouge">Condition</code> blocks printed above, treating each key as single-valued and feeding both policies the identical <code class="language-plaintext highlighter-rouge">aud</code> and <code class="language-plaintext highlighter-rouge">sub</code> values per row. That is a consistency check on the article’s own examples, not an IAM authorization simulation and not a claim about any real policy. The broad policy is shown because it explains row four, not as a recommendation.</em></p> <p>Row four is the whole point: the environment job fails the exact-branch condition and satisfies the repository wildcard, which is how a historical broad policy explains an issuance that today’s narrow policy would refuse. Row seven is the quiet one - a single capitalized character in an organization name breaks both policies, because both operators are case sensitive.</p> <p>Row eight is the abstract case - the key is simply not in the request context - and it is in the table because IAM’s documented answer there is <em>false</em>, not <em>skip this condition</em>. Be precise about what row eight is not, because this is where the subject-format story gets misremembered. A format migration does not remove the <code class="language-plaintext highlighter-rouge">sub</code> key. An immutable-ID subject or a customized template still arrives as a <code class="language-plaintext highlighter-rouge">sub</code> claim; the claim is present, it just carries a value that no longer matches a condition written against the older literal string. That failure has row five’s shape - a value mismatch - not row eight’s. Expect the key, check the value.</p> <p>Which brings up formats. As read today, GitHub documents that <a href="https://docs.github.com/en/actions/reference/security/oidc">repositories created after July 15, 2026 now use an immutable default subject format that includes both the owner ID and repository ID</a>, shaped <code class="language-plaintext highlighter-rouge">repo:OWNER@OWNER-ID/REPO@REPO-ID:ref:refs/heads/BRANCH</code>. Repositories created earlier <a href="https://docs.github.com/en/actions/reference/security/oidc">keep the previous format unless you opt in</a>, and <a href="https://docs.github.com/en/actions/reference/security/oidc">repository renames and transfers after July 15, 2026 also move to the immutable subject format</a>. The rollout <a href="https://docs.github.com/en/actions/reference/security/oidc">does not include GitHub Enterprise Server</a>.</p> <p>Do not turn that into date arithmetic. Establish the format your investigation actually saw from the recorded subject and the dated configuration, because opt-in, renames and transfers all move a repository between formats without changing its creation date. The IDs address namespace reuse - they stop a recycled name from producing a previously-issued subject - and that is all they address. They are not evidence of code integrity or an attribution to a person, the human-readable names are still in the string, and an ID-bearing subject sails straight through a condition wildcarded broadly enough to ignore the segment the IDs live in.</p> <p>Subject customization complicates this further. A template <a href="https://docs.github.com/en/actions/reference/security/oidc">results in a new format for the entire <code class="language-plaintext highlighter-rouge">sub</code> claim</a>, which can carry <a href="https://docs.github.com/en/actions/reference/security/oidc">the value of the <code class="language-plaintext highlighter-rouge">job_workflow_ref</code> claim</a> for enforcing a reusable workflow. If a template was in force, the default rules above do not describe your subject at all. Where immutable IDs apply, GitHub notes they <a href="https://docs.github.com/en/actions/reference/security/oidc">are always included in the <code class="language-plaintext highlighter-rouge">repo</code> segment even when you customize claims</a>.</p> <p>What you cannot do is condition on arbitrary GitHub claims in AWS. The token carries <code class="language-plaintext highlighter-rouge">run_id</code>, <code class="language-plaintext highlighter-rouge">run_attempt</code>, <code class="language-plaintext highlighter-rouge">workflow_ref</code>, <code class="language-plaintext highlighter-rouge">job_workflow_ref</code>, <code class="language-plaintext highlighter-rouge">runner_environment</code> and more as <a href="https://docs.github.com/en/actions/reference/security/oidc">custom claims provided by GitHub</a>, but GitHub’s AWS guide states flatly that <a href="https://docs.github.com/en/actions/how-tos/secure-your-work/security-harden-deployments/oidc-in-aws">support for custom claims for OIDC is unavailable in AWS</a>. If you catch yourself writing <code class="language-plaintext highlighter-rouge">token.actions.githubusercontent.com:run_id</code> into a report, you have invented a condition key.</p> <p>Two closing notes on this layer. IAM does enforce a minimum: when GitHub’s OIDC IdP is the trusted principal, IAM checks at trust-policy creation or update time that <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_create_for-idp_oidc.html">the condition key <code class="language-plaintext highlighter-rouge">token.actions.githubusercontent.com:sub</code> is present and that its value is not solely a wildcard character or null</a>. That is a floor, not evidence of appropriate scoping. And environments are a separate control surface: AWS recommends <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_create_for-idp_oidc.html">using deployment branches and tags to restrict which branches and tags can deploy to the environment</a>. An <code class="language-plaintext highlighter-rouge">environment:prod</code> subject proves the job referenced an environment named <code class="language-plaintext highlighter-rouge">prod</code>. It does not prove a branch restriction or a required review existed at the time, let alone that one was satisfied. Those are separate artifacts: the environment’s historical configuration, and the run’s own approval history, which has a documented read route in <code class="language-plaintext highlighter-rouge">GET /repos/{owner}/{repo}/actions/runs/{run_id}/approvals</code> - <a href="https://docs.github.com/en/rest/actions/workflow-runs">the review history for a workflow run</a>, returning the environments involved, a state, the reviewing user and their comment. An empty review history and an unprotected environment look identical in the subject and nothing alike in the report.</p> <p><br /></p> <p><strong>Separate what the session could do from what it did</strong></p> <p>Successful issuance says AWS accepted the identity. It says nothing about resource permissions. Pick one concrete question: could this session perform <code class="language-plaintext highlighter-rouge">s3:PutObject</code> on <code class="language-plaintext highlighter-rouge">arn:aws:s3:::amzn-s3-demo-bucket/deploy/example.txt</code>?</p> <p>For that you read the role’s permissions, which live in more than one place. Inline documents come from <code class="language-plaintext highlighter-rouge">GetRolePolicy</code>; for attached managed policies you <a href="https://docs.aws.amazon.com/cli/latest/reference/iam/get-role-policy.html">use <code class="language-plaintext highlighter-rouge">GetPolicy</code> to determine the policy’s default version, then use <code class="language-plaintext highlighter-rouge">GetPolicyVersion</code> to retrieve the policy document</a>. Getting that two-step wrong is a common way to read the wrong document entirely.</p> <div class="language-json highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="p">{</span><span class="w"> </span><span class="nl">"Version"</span><span class="p">:</span><span class="w"> </span><span class="s2">"2012-10-17"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Statement"</span><span class="p">:</span><span class="w"> </span><span class="p">[</span><span class="w"> </span><span class="p">{</span><span class="w"> </span><span class="nl">"Sid"</span><span class="p">:</span><span class="w"> </span><span class="s2">"WriteDeployPrefix"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Effect"</span><span class="p">:</span><span class="w"> </span><span class="s2">"Allow"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Action"</span><span class="p">:</span><span class="w"> </span><span class="s2">"s3:PutObject"</span><span class="p">,</span><span class="w"> </span><span class="nl">"Resource"</span><span class="p">:</span><span class="w"> </span><span class="s2">"arn:aws:s3:::amzn-s3-demo-bucket/deploy/*"</span><span class="w"> </span><span class="p">}</span><span class="w"> </span><span class="p">]</span><span class="w"> </span><span class="p">}</span><span class="w"> </span></code></pre></div></div> <p>Versioning has its own temporal trap, symmetrical to the trust-policy one. A managed policy’s default version is <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_managed-versioning.html">the operative version</a>, a managed policy <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_managed-versioning.html">can have up to five versions</a>, older versions can be deleted, and <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_managed-versioning.html">IAM does not support versioning for inline policies</a> at all. So today’s default version is not necessarily April’s default version, and for an inline policy there is no revision history inside IAM to walk - the same fact about IAM, with the same external alternatives, as in the trust-policy section. Track <code class="language-plaintext highlighter-rouge">CreatePolicyVersion</code> and <code class="language-plaintext highlighter-rouge">SetDefaultPolicyVersion</code> events with their outcomes. And the <code class="language-plaintext highlighter-rouge">"Version": "2012-10-17"</code> string at the top of every document is the policy language version, not a revision date - it will be identical on documents written years apart.</p> <p>The evaluation itself is not a single lookup. <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_evaluation-logic_policy-eval-denyallow.html">By default, all requests are implicitly denied</a>, requests <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_evaluation-logic_policy-eval-denyallow.html">must be explicitly allowed by a policy or set of policies</a>, and <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_evaluation-logic_policy-eval-denyallow.html">an explicit deny overrides an explicit allow</a>. The policies in scope for the deny sweep are broad: <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_evaluation-logic_policy-eval-denyallow.html">AWS Organizations SCPs and RCPs, resource-based policies, identity-based policies, IAM permissions boundaries, and session policies</a>. Session policies matter directly here, because <a href="https://docs.aws.amazon.com/STS/latest/APIReference/API_AssumeRoleWithWebIdentity.html">passing policies to this operation returns new temporary credentials whose permissions are the intersection of the role’s identity-based policy and the session policies</a>, and the AWS credentials action exposes both <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">inline and managed session policy inputs</a>.</p> <p>But do not reach for “effective permissions are the intersection of everything” as a universal rule, because there is a documented exception that runs the other way. Within the same account, a resource-based policy that grants permissions to an IAM role <em>session</em> ARN grants them directly to that session, and <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/reference_policies_evaluation-logic_policy-eval-denyallow.html">permissions granted directly to a session are not limited by an implicit deny in an identity-based policy, a permissions boundary, or session policy</a>. A bucket policy naming <code class="language-plaintext highlighter-rouge">arn:aws:sts::111122223333:assumed-role/example-deploy-role/700001-2</code> can therefore allow something the role’s own identity policy never mentions. That exception covers implicit denies only. Explicit denies and applicable organization controls still bite.</p> <p>So you get three honest verdicts, not one. Before the first one can mean anything, though, I have to say what this synthetic case stipulates. The moment that matters is when AWS evaluated the request, which is not when the credentials were issued - and <code class="language-plaintext highlighter-rouge">09:13:22</code> does not name that moment either, because it is a completion stamp. It closes the interval rather than marking a point inside it. So the stipulation has to cover that whole interval, from the request through the completion it records:</p> <ul> <li>the identity policy fragment above was attached to the role and was the operative document then;</li> <li>the bucket is a same-account general purpose bucket using SSE-S3, and the write is one ordinary object;</li> <li>no session policy was passed on the assumption, and no permissions boundary applied to the role;</li> <li>any applicable SCPs and RCPs permitted the action - a stipulation about this invented account, not a suggestion that organizations generally lack such controls;</li> <li>no resource-based policy or other applicable policy carried an explicit deny reaching this action and resource;</li> <li>no KMS, ACL, tagging, Object Lock, access point or VPC endpoint condition was in play, and no bucket policy granted anything directly to the session ARN.</li> </ul> <p>Those six lines are stipulations for an illustration, asserted across that interval rather than at an instant. They are not retained facts, and the reason to write them out rather than wave at them is that each one is a document you either hold or do not:</p> <ul> <li><strong>Allowed under stated assumptions.</strong> <code class="language-plaintext highlighter-rouge">s3:PutObject</code> on <code class="language-plaintext highlighter-rouge">arn:aws:s3:::amzn-s3-demo-bucket/deploy/example.txt</code> is permitted given that baseline. The named object, not the prefix - the allow statement is wider than the object, but your verdict should not inherit that width, because every object you fold into it is one more request context you are silently asserting the baseline held for.</li> <li><strong>Denied.</strong> An applicable explicit deny covers the action and resource: the session was admitted to AWS and still could not do this thing. Issuance and authorization are sequential stages with different inputs and different policy documents behind them, not one decision taken twice.</li> <li><strong>Undetermined.</strong> You have the trust policy and the issuance, but a material policy document from the relevant time is unavailable. Undetermined is a real finding and belongs in the report as one. Reconstructing the missing policy from whatever would explain the outcome is circular reasoning in a lab coat.</li> </ul> <p>Then, separately, look at what actually happened - and notice that the excerpt printed earlier cannot answer that by itself. An allowlisted excerpt with no error fields is not the same artifact as a complete record with no error fields; the first is equally consistent with my having dropped them on the way in. I checked the underlying synthetic record before excerpting it, found no error fields, and this example stipulates that the operation succeeded. That is a statement about a fixture I built; in your case it is a check you run against the complete record, before an outcome goes anywhere near the report.</p> <p>Where to look. Errors appear in the record’s <code class="language-plaintext highlighter-rouge">errorCode</code>, <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-record-contents.html">the AWS service error if the request returns an error</a>, and some services put <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/cloudtrail-event-reference-record-contents.html">error information as part of <code class="language-plaintext highlighter-rouge">responseElements</code></a> instead. The standard record reference does not define a general HTTP status field carried by every event, so do not invent one and do not read the absence of error fields as a recorded HTTP 200. That is a statement about the common record shape: where a particular service does surface status in its own records, cite it. Once you do have a complete record with no reported error, S3’s own semantics carry you a useful distance - <a href="https://docs.aws.amazon.com/AmazonS3/latest/API/API_PutObject.html">if the action is successful, the service sends back an HTTP 200 response</a>, and <a href="https://docs.aws.amazon.com/AmazonS3/latest/API/API_PutObject.html">Amazon S3 never adds partial objects; if you receive a success response, Amazon S3 added the entire object to the bucket</a>. So a clean record supports “the object write completed”, not “something half-landed in the bucket”.</p> <p>And keep the ladder straight. What the policy permitted, what the API returned, whether the deployment was approved, and whether the object’s contents were benign are four separate claims. A complete, error-free record settles the second, within the limits of what was collected, and says nothing about the third or fourth.</p> <p><br /></p> <p><strong>Establish coverage before you claim absence</strong></p> <p>Every “we found no evidence of X” is really a statement about your collection. Get the coverage story written down before the conclusions.</p> <p>The console’s Event history is the narrowest surface. AWS states that <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/view-cloudtrail-events.html">an event history search is limited to a single AWS account, only returns events from a single AWS Region, and cannot query multiple attributes</a>, and that <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/view-cloudtrail-events.html">it does not show data events, Insights events, or network activity events</a>. The API behind it is bounded the same way: you can <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">look up events that occurred in a Region within the last 90 days</a>, and the attribute list <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">can contain only one item</a>.</p> <p>That 90-day bound is the first thing to check against your own dates, and here it decides the shape of the entire section. The issuance is dated 2026-04-14. Ninety days ran out in July, so no lookup run today reaches it and no parameter on the command fixes that. What follows is therefore the command the responder on duty ran on 2026-04-15, while the event was still inside the window - writing the response into the case’s evidence store, where it has been sitting ever since.</p> <div class="language-bash highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="c"># Run 2026-04-15, while the event was still inside the 90-day window.</span> <span class="c"># Pre-provisioned read-only profile. No credential or sign-in commands here.</span> aws cloudtrail lookup-events <span class="se">\</span> <span class="nt">--profile</span> ir-readonly-111122223333 <span class="se">\</span> <span class="nt">--region</span> us-west-2 <span class="se">\</span> <span class="nt">--lookup-attributes</span> <span class="nv">AttributeKey</span><span class="o">=</span>EventName,AttributeValue<span class="o">=</span>AssumeRoleWithWebIdentity <span class="se">\</span> <span class="nt">--start-time</span> 2026-04-14T00:00:00Z <span class="se">\</span> <span class="nt">--end-time</span> 2026-04-15T00:00:00Z <span class="se">\</span> <span class="nt">--output</span> json <span class="se">\</span> <span class="o">&gt;</span> /cases/2026-0414-deploy-role/evidence/sts-lookup-us-west-2.json </code></pre></div></div> <p><em>Editorial Note: printed for reading, not for running, and the profile name is invented. The point is that the command names a deliberate, recorded configuration instead of inheriting whatever the shell happened to hold - not that the string proves anything about the account behind it. A profile authorizes exactly what the credentials behind it authorize; it is not standing permission to sweep an organization, and whose account you may read is a question for your own authorization process.</em></p> <p>Inheriting an old case means inheriting that file, the trail’s S3 objects, or a CloudTrail Lake event data store whose retention still covers April. No query recovers activity nobody recorded.</p> <p><code class="language-plaintext highlighter-rouge">--profile</code> selects a named configuration explicitly, so the capture records a deliberate choice rather than ambient credentials - but a profile name is a label, not a binding. Nothing about the string <code class="language-plaintext highlighter-rouge">ir-readonly-111122223333</code> establishes that the profile resolves to account <code class="language-plaintext highlighter-rouge">111122223333</code>, or that it is read-only. Confirm which account you are actually reading, under your own authorization, and write down what you confirmed.</p> <p><code class="language-plaintext highlighter-rouge">--region</code> is the one that genuinely pins something: it selects the Region explicitly and <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">overrides config and environment settings</a>, because a capture that does not say which Region it came from is not evidence. The time bounds are inclusive at both ends - <code class="language-plaintext highlighter-rouge">--start-time</code> returns <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">only events that occur after or at the specified time</a>, <code class="language-plaintext highlighter-rouge">--end-time</code> <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">only events that occur before or at the specified time</a> - and they are written in UTC to match <code class="language-plaintext highlighter-rouge">eventTime</code>. There is exactly one lookup attribute, because one is all the API accepts.</p> <p>There is deliberately no <code class="language-plaintext highlighter-rouge">--no-paginate</code> and no <code class="language-plaintext highlighter-rouge">--max-items</code>: <code class="language-plaintext highlighter-rouge">lookup-events</code> is a paginated operation and <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">multiple API calls may be issued in order to retrieve the entire data set of results</a>, so leaving the CLI’s automatic pagination alone is what gets you the whole set rather than the first slice. If you have seen the number 50 attached to this API, that is the service’s page size, not a result cap: AWS describes <code class="language-plaintext highlighter-rouge">--page-size</code> as <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">the size of each page to get in the AWS service call, which does not affect the number of items returned in the command’s output</a>. Do not report fifty as a ceiling.</p> <p>Redirecting to a file is deliberate for the reason given earlier: the response carries full STS <code class="language-plaintext highlighter-rouge">responseElements</code>. It belongs in the case’s restricted evidence directory - not a repository, not a synced drive, not terminal scrollback - and that exact path is the input the filter below opens.</p> <p>Note also what this command is not for: it looks up management events, so the <code class="language-plaintext highlighter-rouge">PutObject</code> from earlier will never appear in its output. That is the second coverage trap below.</p> <p>Because you can only filter on one attribute, narrowing to a specific role happens locally - and it happens on parsed structure, not on text. The <code class="language-plaintext highlighter-rouge">CloudTrailEvent</code> field is <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">a JSON string that contains a representation of the event returned</a>, so it arrives as an escaped blob. Grepping that blob for an ARN matches role names inside unrelated fields and misses anything the escaping mangles.</p> <p>Most of the filter below is not filtering. It is the difference between <em>no matching events</em> and <em>this input cannot tell me whether there were matching events</em> - the one distinction in this article that a script can quietly get wrong on your behalf.</p> <div class="language-python highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="kn">import</span> <span class="nn">json</span> <span class="kn">import</span> <span class="nn">sys</span> <span class="n">ROLE</span> <span class="o">=</span> <span class="s">"arn:aws:iam::111122223333:role/example-deploy-role"</span> <span class="n">CAPTURE</span> <span class="o">=</span> <span class="s">"/cases/2026-0414-deploy-role/evidence/sts-lookup-us-west-2.json"</span> <span class="n">ABSENT</span> <span class="o">=</span> <span class="s">"-"</span> <span class="n">MALFORMED</span> <span class="o">=</span> <span class="s">"&lt;malformed&gt;"</span> <span class="k">def</span> <span class="nf">usable</span><span class="p">(</span><span class="n">value</span><span class="p">):</span> <span class="s">"""A consumed field is usable only as a non-blank string."""</span> <span class="k">return</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">value</span><span class="p">,</span> <span class="nb">str</span><span class="p">)</span> <span class="ow">and</span> <span class="nb">bool</span><span class="p">(</span><span class="n">value</span><span class="p">.</span><span class="n">strip</span><span class="p">())</span> <span class="k">def</span> <span class="nf">render</span><span class="p">(</span><span class="n">value</span><span class="p">):</span> <span class="s">"""Escape one consumed value for printing. Returns (text, malformed)."""</span> <span class="k">if</span> <span class="n">value</span> <span class="ow">is</span> <span class="bp">None</span><span class="p">:</span> <span class="k">return</span> <span class="n">ABSENT</span><span class="p">,</span> <span class="bp">False</span> <span class="k">if</span> <span class="ow">not</span> <span class="n">usable</span><span class="p">(</span><span class="n">value</span><span class="p">):</span> <span class="k">return</span> <span class="n">MALFORMED</span><span class="p">,</span> <span class="bp">True</span> <span class="k">return</span> <span class="n">json</span><span class="p">.</span><span class="n">dumps</span><span class="p">(</span><span class="n">value</span><span class="p">,</span> <span class="n">ensure_ascii</span><span class="o">=</span><span class="bp">True</span><span class="p">),</span> <span class="bp">False</span> <span class="k">def</span> <span class="nf">classify</span><span class="p">(</span><span class="n">item</span><span class="p">):</span> <span class="s">"""Bucket one Events row. Returns (verdict, reason, record)."""</span> <span class="k">if</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">item</span><span class="p">,</span> <span class="nb">dict</span><span class="p">):</span> <span class="k">return</span> <span class="s">"unreadable"</span><span class="p">,</span> <span class="s">"row is not an object"</span><span class="p">,</span> <span class="bp">None</span> <span class="n">blob</span> <span class="o">=</span> <span class="n">item</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"CloudTrailEvent"</span><span class="p">)</span> <span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">blob</span><span class="p">,</span> <span class="nb">dict</span><span class="p">):</span> <span class="n">record</span> <span class="o">=</span> <span class="n">blob</span> <span class="k">elif</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">blob</span><span class="p">,</span> <span class="nb">str</span><span class="p">):</span> <span class="k">try</span><span class="p">:</span> <span class="n">record</span> <span class="o">=</span> <span class="n">json</span><span class="p">.</span><span class="n">loads</span><span class="p">(</span><span class="n">blob</span><span class="p">)</span> <span class="k">except</span> <span class="nb">ValueError</span><span class="p">:</span> <span class="k">return</span> <span class="s">"unreadable"</span><span class="p">,</span> <span class="s">"CloudTrailEvent is not valid JSON"</span><span class="p">,</span> <span class="bp">None</span> <span class="k">else</span><span class="p">:</span> <span class="k">return</span> <span class="s">"unreadable"</span><span class="p">,</span> <span class="s">"CloudTrailEvent missing or not a JSON string"</span><span class="p">,</span> <span class="bp">None</span> <span class="k">if</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">record</span><span class="p">,</span> <span class="nb">dict</span><span class="p">):</span> <span class="k">return</span> <span class="s">"unreadable"</span><span class="p">,</span> <span class="s">"CloudTrailEvent did not decode to an object"</span><span class="p">,</span> <span class="bp">None</span> <span class="n">source</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"eventSource"</span><span class="p">)</span> <span class="n">name</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"eventName"</span><span class="p">)</span> <span class="k">if</span> <span class="ow">not</span> <span class="n">usable</span><span class="p">(</span><span class="n">source</span><span class="p">)</span> <span class="ow">or</span> <span class="ow">not</span> <span class="n">usable</span><span class="p">(</span><span class="n">name</span><span class="p">):</span> <span class="k">return</span> <span class="p">(</span><span class="s">"indeterminate"</span><span class="p">,</span> <span class="s">"eventSource or eventName absent, blank or not a string"</span><span class="p">,</span> <span class="n">record</span><span class="p">)</span> <span class="k">if</span> <span class="n">source</span> <span class="o">!=</span> <span class="s">"sts.amazonaws.com"</span> <span class="ow">or</span> <span class="n">name</span> <span class="o">!=</span> <span class="s">"AssumeRoleWithWebIdentity"</span><span class="p">:</span> <span class="k">return</span> <span class="s">"out_of_scope"</span><span class="p">,</span> <span class="s">"a different well-formed service or operation"</span><span class="p">,</span> <span class="n">record</span> <span class="n">params</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"requestParameters"</span><span class="p">)</span> <span class="k">if</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">params</span><span class="p">,</span> <span class="nb">dict</span><span class="p">):</span> <span class="k">return</span> <span class="s">"indeterminate"</span><span class="p">,</span> <span class="s">"requestParameters absent or not an object"</span><span class="p">,</span> <span class="n">record</span> <span class="n">role_arn</span> <span class="o">=</span> <span class="n">params</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"roleArn"</span><span class="p">)</span> <span class="k">if</span> <span class="ow">not</span> <span class="n">usable</span><span class="p">(</span><span class="n">role_arn</span><span class="p">):</span> <span class="k">return</span> <span class="s">"indeterminate"</span><span class="p">,</span> <span class="s">"roleArn absent, blank or not a string"</span><span class="p">,</span> <span class="n">record</span> <span class="k">if</span> <span class="n">role_arn</span> <span class="o">!=</span> <span class="n">ROLE</span><span class="p">:</span> <span class="k">return</span> <span class="s">"other_role"</span><span class="p">,</span> <span class="s">"roleArn names a different role"</span><span class="p">,</span> <span class="n">record</span> <span class="k">return</span> <span class="s">"match"</span><span class="p">,</span> <span class="s">""</span><span class="p">,</span> <span class="n">record</span> <span class="k">def</span> <span class="nf">excerpt</span><span class="p">(</span><span class="n">record</span><span class="p">):</span> <span class="s">"""Output allowlist. Never widen this to the whole responseElements object."""</span> <span class="n">response</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"responseElements"</span><span class="p">)</span> <span class="n">response</span> <span class="o">=</span> <span class="n">response</span> <span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">response</span><span class="p">,</span> <span class="nb">dict</span><span class="p">)</span> <span class="k">else</span> <span class="p">{}</span> <span class="n">creds</span> <span class="o">=</span> <span class="n">response</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"credentials"</span><span class="p">)</span> <span class="n">creds</span> <span class="o">=</span> <span class="n">creds</span> <span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">creds</span><span class="p">,</span> <span class="nb">dict</span><span class="p">)</span> <span class="k">else</span> <span class="p">{}</span> <span class="n">raw_key</span> <span class="o">=</span> <span class="n">creds</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"accessKeyId"</span><span class="p">)</span> <span class="n">raw_code</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"errorCode"</span><span class="p">)</span> <span class="n">has_message</span> <span class="o">=</span> <span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"errorMessage"</span><span class="p">)</span> <span class="ow">is</span> <span class="ow">not</span> <span class="bp">None</span> <span class="k">if</span> <span class="n">raw_key</span> <span class="ow">is</span> <span class="bp">None</span><span class="p">:</span> <span class="n">key_state</span> <span class="o">=</span> <span class="s">"absent"</span> <span class="k">else</span><span class="p">:</span> <span class="n">key_state</span> <span class="o">=</span> <span class="s">"present"</span> <span class="k">if</span> <span class="n">usable</span><span class="p">(</span><span class="n">raw_key</span><span class="p">)</span> <span class="k">else</span> <span class="s">"malformed"</span> <span class="k">if</span> <span class="n">raw_code</span> <span class="ow">is</span> <span class="bp">None</span><span class="p">:</span> <span class="c1"># A message with no code is an error signal this script cannot read. </span> <span class="n">code_state</span> <span class="o">=</span> <span class="s">"malformed"</span> <span class="k">if</span> <span class="n">has_message</span> <span class="k">else</span> <span class="s">"unreported"</span> <span class="k">else</span><span class="p">:</span> <span class="n">code_state</span> <span class="o">=</span> <span class="s">"reported"</span> <span class="k">if</span> <span class="n">usable</span><span class="p">(</span><span class="n">raw_code</span><span class="p">)</span> <span class="k">else</span> <span class="s">"malformed"</span> <span class="n">conflict</span> <span class="o">=</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"reported"</span> <span class="ow">and</span> <span class="n">key_state</span> <span class="o">==</span> <span class="s">"present"</span> <span class="k">if</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"malformed"</span> <span class="ow">or</span> <span class="n">key_state</span> <span class="o">==</span> <span class="s">"malformed"</span> <span class="ow">or</span> <span class="n">conflict</span><span class="p">:</span> <span class="n">outcome</span> <span class="o">=</span> <span class="s">"outcome-unclear"</span> <span class="k">elif</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"reported"</span><span class="p">:</span> <span class="n">outcome</span> <span class="o">=</span> <span class="s">"rejected"</span> <span class="k">elif</span> <span class="n">key_state</span> <span class="o">==</span> <span class="s">"present"</span><span class="p">:</span> <span class="n">outcome</span> <span class="o">=</span> <span class="s">"issued"</span> <span class="k">else</span><span class="p">:</span> <span class="n">outcome</span> <span class="o">=</span> <span class="s">"outcome-unclear"</span> <span class="n">event_time</span><span class="p">,</span> <span class="n">bad_time</span> <span class="o">=</span> <span class="n">render</span><span class="p">(</span><span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"eventTime"</span><span class="p">))</span> <span class="n">event_id</span><span class="p">,</span> <span class="n">bad_id</span> <span class="o">=</span> <span class="n">render</span><span class="p">(</span><span class="n">record</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"eventID"</span><span class="p">))</span> <span class="n">subject</span><span class="p">,</span> <span class="n">bad_subject</span> <span class="o">=</span> <span class="n">render</span><span class="p">(</span><span class="n">response</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"subjectFromWebIdentityToken"</span><span class="p">))</span> <span class="n">key</span><span class="p">,</span> <span class="n">_</span> <span class="o">=</span> <span class="n">render</span><span class="p">(</span><span class="n">raw_key</span><span class="p">)</span> <span class="k">if</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"reported"</span><span class="p">:</span> <span class="n">code</span> <span class="o">=</span> <span class="n">json</span><span class="p">.</span><span class="n">dumps</span><span class="p">(</span><span class="n">raw_code</span><span class="p">,</span> <span class="n">ensure_ascii</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span> <span class="k">elif</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"malformed"</span><span class="p">:</span> <span class="n">code</span> <span class="o">=</span> <span class="n">MALFORMED</span> <span class="k">else</span><span class="p">:</span> <span class="n">code</span> <span class="o">=</span> <span class="n">ABSENT</span> <span class="k">return</span> <span class="p">{</span> <span class="s">"eventTime"</span><span class="p">:</span> <span class="n">event_time</span><span class="p">,</span> <span class="s">"eventID"</span><span class="p">:</span> <span class="n">event_id</span><span class="p">,</span> <span class="s">"errorCode"</span><span class="p">:</span> <span class="n">code</span><span class="p">,</span> <span class="s">"subject"</span><span class="p">:</span> <span class="n">subject</span><span class="p">,</span> <span class="s">"accessKeyId"</span><span class="p">:</span> <span class="n">key</span><span class="p">,</span> <span class="s">"outcome"</span><span class="p">:</span> <span class="n">outcome</span><span class="p">,</span> <span class="s">"unresolved"</span><span class="p">:</span> <span class="nb">any</span><span class="p">((</span><span class="n">bad_time</span><span class="p">,</span> <span class="n">bad_id</span><span class="p">,</span> <span class="n">bad_subject</span><span class="p">,</span> <span class="n">conflict</span><span class="p">,</span> <span class="n">key_state</span> <span class="o">==</span> <span class="s">"malformed"</span><span class="p">,</span> <span class="n">code_state</span> <span class="o">==</span> <span class="s">"malformed"</span><span class="p">)),</span> <span class="p">}</span> <span class="k">try</span><span class="p">:</span> <span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">CAPTURE</span><span class="p">,</span> <span class="n">encoding</span><span class="o">=</span><span class="s">"utf-8"</span><span class="p">)</span> <span class="k">as</span> <span class="n">fh</span><span class="p">:</span> <span class="n">payload</span> <span class="o">=</span> <span class="n">json</span><span class="p">.</span><span class="n">load</span><span class="p">(</span><span class="n">fh</span><span class="p">)</span> <span class="k">except</span> <span class="p">(</span><span class="nb">OSError</span><span class="p">,</span> <span class="nb">ValueError</span><span class="p">)</span> <span class="k">as</span> <span class="n">exc</span><span class="p">:</span> <span class="c1"># Type only. A path or parse message can echo capture contents into a log. </span> <span class="k">print</span><span class="p">(</span><span class="s">"capture unusable ({}); no coverage claim rests on this run"</span><span class="p">.</span><span class="nb">format</span><span class="p">(</span> <span class="nb">type</span><span class="p">(</span><span class="n">exc</span><span class="p">).</span><span class="n">__name__</span><span class="p">),</span> <span class="nb">file</span><span class="o">=</span><span class="n">sys</span><span class="p">.</span><span class="n">stderr</span><span class="p">)</span> <span class="n">sys</span><span class="p">.</span><span class="nb">exit</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span> <span class="k">if</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">payload</span><span class="p">,</span> <span class="nb">dict</span><span class="p">)</span> <span class="ow">or</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">payload</span><span class="p">.</span><span class="n">get</span><span class="p">(</span><span class="s">"Events"</span><span class="p">),</span> <span class="nb">list</span><span class="p">):</span> <span class="k">print</span><span class="p">(</span><span class="s">"capture is not a LookupEvents response envelope"</span><span class="p">,</span> <span class="nb">file</span><span class="o">=</span><span class="n">sys</span><span class="p">.</span><span class="n">stderr</span><span class="p">)</span> <span class="n">sys</span><span class="p">.</span><span class="nb">exit</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span> <span class="n">counts</span> <span class="o">=</span> <span class="p">{</span><span class="s">"match"</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s">"other_role"</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s">"out_of_scope"</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s">"indeterminate"</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s">"unreadable"</span><span class="p">:</span> <span class="mi">0</span><span class="p">}</span> <span class="n">unresolved</span> <span class="o">=</span> <span class="mi">0</span> <span class="k">for</span> <span class="n">index</span><span class="p">,</span> <span class="n">item</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">payload</span><span class="p">[</span><span class="s">"Events"</span><span class="p">]):</span> <span class="n">verdict</span><span class="p">,</span> <span class="n">reason</span><span class="p">,</span> <span class="n">record</span> <span class="o">=</span> <span class="n">classify</span><span class="p">(</span><span class="n">item</span><span class="p">)</span> <span class="n">counts</span><span class="p">[</span><span class="n">verdict</span><span class="p">]</span> <span class="o">+=</span> <span class="mi">1</span> <span class="k">if</span> <span class="n">verdict</span> <span class="o">==</span> <span class="s">"match"</span><span class="p">:</span> <span class="n">row</span> <span class="o">=</span> <span class="n">excerpt</span><span class="p">(</span><span class="n">record</span><span class="p">)</span> <span class="k">if</span> <span class="n">row</span><span class="p">[</span><span class="s">"unresolved"</span><span class="p">]:</span> <span class="n">unresolved</span> <span class="o">+=</span> <span class="mi">1</span> <span class="k">print</span><span class="p">(</span><span class="s">"row {} match {} {} {} sub={} key={} err={}"</span><span class="p">.</span><span class="nb">format</span><span class="p">(</span> <span class="n">index</span><span class="p">,</span> <span class="n">row</span><span class="p">[</span><span class="s">"eventTime"</span><span class="p">],</span> <span class="n">row</span><span class="p">[</span><span class="s">"eventID"</span><span class="p">],</span> <span class="n">row</span><span class="p">[</span><span class="s">"outcome"</span><span class="p">],</span> <span class="n">row</span><span class="p">[</span><span class="s">"subject"</span><span class="p">],</span> <span class="n">row</span><span class="p">[</span><span class="s">"accessKeyId"</span><span class="p">],</span> <span class="n">row</span><span class="p">[</span><span class="s">"errorCode"</span><span class="p">]))</span> <span class="k">elif</span> <span class="n">verdict</span> <span class="ow">in</span> <span class="p">(</span><span class="s">"indeterminate"</span><span class="p">,</span> <span class="s">"unreadable"</span><span class="p">):</span> <span class="k">print</span><span class="p">(</span><span class="s">"row {} {}: {}"</span><span class="p">.</span><span class="nb">format</span><span class="p">(</span><span class="n">index</span><span class="p">,</span> <span class="n">verdict</span><span class="p">,</span> <span class="n">reason</span><span class="p">),</span> <span class="nb">file</span><span class="o">=</span><span class="n">sys</span><span class="p">.</span><span class="n">stderr</span><span class="p">)</span> <span class="k">print</span><span class="p">(</span><span class="s">"rows={} match={} other_role={} out_of_scope={} "</span> <span class="s">"indeterminate={} unreadable={} unresolved={}"</span><span class="p">.</span><span class="nb">format</span><span class="p">(</span> <span class="nb">len</span><span class="p">(</span><span class="n">payload</span><span class="p">[</span><span class="s">"Events"</span><span class="p">]),</span> <span class="n">counts</span><span class="p">[</span><span class="s">"match"</span><span class="p">],</span> <span class="n">counts</span><span class="p">[</span><span class="s">"other_role"</span><span class="p">],</span> <span class="n">counts</span><span class="p">[</span><span class="s">"out_of_scope"</span><span class="p">],</span> <span class="n">counts</span><span class="p">[</span><span class="s">"indeterminate"</span><span class="p">],</span> <span class="n">counts</span><span class="p">[</span><span class="s">"unreadable"</span><span class="p">],</span> <span class="n">unresolved</span><span class="p">))</span> <span class="n">incomplete</span> <span class="o">=</span> <span class="nb">bool</span><span class="p">(</span><span class="n">counts</span><span class="p">[</span><span class="s">"indeterminate"</span><span class="p">]</span> <span class="ow">or</span> <span class="n">counts</span><span class="p">[</span><span class="s">"unreadable"</span><span class="p">]</span> <span class="ow">or</span> <span class="n">unresolved</span><span class="p">)</span> <span class="k">if</span> <span class="s">"NextToken"</span> <span class="ow">in</span> <span class="n">payload</span><span class="p">:</span> <span class="k">if</span> <span class="n">usable</span><span class="p">(</span><span class="n">payload</span><span class="p">[</span><span class="s">"NextToken"</span><span class="p">]):</span> <span class="k">print</span><span class="p">(</span><span class="s">"PARTIAL: NextToken set, further pages were not collected"</span><span class="p">,</span> <span class="nb">file</span><span class="o">=</span><span class="n">sys</span><span class="p">.</span><span class="n">stderr</span><span class="p">)</span> <span class="k">else</span><span class="p">:</span> <span class="k">print</span><span class="p">(</span><span class="s">"PARTIAL: NextToken present but not a usable page token; "</span> <span class="s">"completeness is unconfirmed, not disproved"</span><span class="p">,</span> <span class="nb">file</span><span class="o">=</span><span class="n">sys</span><span class="p">.</span><span class="n">stderr</span><span class="p">)</span> <span class="n">incomplete</span> <span class="o">=</span> <span class="bp">True</span> <span class="n">sys</span><span class="p">.</span><span class="nb">exit</span><span class="p">(</span><span class="mi">1</span> <span class="k">if</span> <span class="n">incomplete</span> <span class="k">else</span> <span class="mi">0</span><span class="p">)</span> </code></pre></div></div> <p><em>Editorial Note: this parses JSON and compares strings. It is not an IAM decision, not a schema validation, and it contacts nothing. While writing this section I ran the exact block above offline against 50 synthetic in-memory inputs - well-formed rows, wrong-service and wrong-role rows, absent and wrong-typed discriminators, blank and wrong-typed role ARNs, malformed envelopes and rows, nested objects planted in the fields it prints, conflicting error-and-credential combinations, and several <code class="language-plaintext highlighter-rouge">NextToken</code> shapes - with the file read faked and both output streams captured. That establishes how this code behaves on those inputs and nothing whatever about AWS.</em></p> <p>Four deliberate choices in there, all of them about refusing to round an unknown down to a known. It validates <code class="language-plaintext highlighter-rouge">eventSource</code> and <code class="language-plaintext highlighter-rouge">eventName</code> as non-blank strings before comparing them, so a row whose discriminators are missing, null or some other type becomes indeterminate rather than a confident exclusion - while a row naming a genuinely different service or operation stays an ordinary out-of-scope miss. It keeps <em>this record does not concern your role</em> separate from <em>this record cannot tell me either way</em>, so an absent, blank or wrong-typed <code class="language-plaintext highlighter-rouge">roleArn</code> is indeterminate, and one malformed row does not silently take the rest of the file down with it. It type-checks and escapes every value it prints, so a nested object where a timestamp should be prints as <code class="language-plaintext highlighter-rouge">&lt;malformed&gt;</code> rather than stringifying a credential object into your notes, and a value carrying newlines cannot forge a second output row. And it reads <code class="language-plaintext highlighter-rouge">NextToken</code> by presence and shape rather than truthiness: a present, non-blank string means pages you did not collect, because AWS is explicit that <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">if the token does not appear, there are no more results to return</a>, while a present token of any other shape means completeness is unconfirmed rather than disproved.</p> <p>The output is an allowlist, same discipline as the case notes: event time, event ID, outcome, subject, access key ID, error code, each one rendered as a fixed indicator or an escaped string and never as a stringified container. Failure messages are fixed strings and never echo record, message or exception content. The outcome handling makes four distinctions: a usable key with no reported error is issuance evidence in this record’s documented context; a reported error with no key is a recorded rejection; a key and an error together, a malformed error field, or an <code class="language-plaintext highlighter-rouge">errorMessage</code> with no <code class="language-plaintext highlighter-rouge">errorCode</code> is unclear rather than either; and an absent key stays unclear even when the rest of the record looks healthy, because AWS documents that the field can simply be missing. Exit zero means no unresolved condition in the rows this capture held, under those checks - not that retention covered the period, not that the selectors collected what you needed, and not that the printed fields are proof against every hostile input.</p> <p>Three coverage traps specific to this investigation:</p> <ul> <li><strong><code class="language-plaintext highlighter-rouge">AccessKeyId</code> as a lookup attribute is not your session key.</strong> The lookup output’s <code class="language-plaintext highlighter-rouge">AccessKeyId</code> is <a href="https://docs.aws.amazon.com/cli/latest/reference/cloudtrail/lookup-events.html">the access key ID that was used to sign the request</a>. On the issuance event, the session key lives <em>inside</em> the response, and the request itself is unsigned federation traffic. Use that attribute to find what the session went on to do; do not expect it to surface the issuance.</li> <li><strong>S3 object activity is not in management events at all.</strong> Object operations such as <code class="language-plaintext highlighter-rouge">GetObject</code>, <code class="language-plaintext highlighter-rouge">DeleteObject</code> and <code class="language-plaintext highlighter-rouge">PutObject</code> on general purpose buckets are <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/logging-data-events-with-cloudtrail.html">data events</a>, and <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/logging-data-events-with-cloudtrail.html">by default, trails and event data stores do not log data events</a>. Whether your April write was recorded depends on the selectors in force in April: AWS’s own example walks through an event that <a href="https://docs.aws.amazon.com/awscloudtrail/latest/userguide/logging-data-events-with-cloudtrail.html">occurred on an object that doesn’t match the prefix specified in your trail or event data store</a> and is therefore not logged, and read-only object APIs are dropped by a write-only selector. Check the historical selector configuration, prefix, read/write choice, region and retention before you interpret silence. Turning collection on now recovers nothing from then.</li> <li><strong>Look in the right region for STS.</strong> AWS notes that <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">CloudTrail logs for requests made to the AWS STS global endpoint will be sent to the US East (N. Virginia) Region</a>, while <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">logs for requests served by AWS STS Regional endpoints will continue to be logged to their respective Region</a>. This article’s example stipulates a regional endpoint, hence <code class="language-plaintext highlighter-rouge">us-west-2</code>. Do not assume that; where present, <code class="language-plaintext highlighter-rouge">additionalEventData.RequestDetails.endpointType</code> and <code class="language-plaintext highlighter-rouge">awsServingRegion</code> give you the endpoint type and the serving region, and the serving region is not the logging region.</li> </ul> <p>Last, the hardest one. A missing issuance event is not a documented rejection. CloudTrail logs non-authenticated requests to <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code>, but AWS states that <a href="https://docs.aws.amazon.com/IAM/latest/UserGuide/cloudtrail-integration.html">some non-authenticated AWS STS requests might not be logged because they do not meet the minimum expectation of being sufficiently valid to be trusted as a legitimate request</a>. No event is a coverage statement, not proof that a request was denied or that nothing was attempted.</p> <p><br /></p> <p><strong>Write the scope statement in layers</strong></p> <p>The deliverable is not “the workflow was fine” or “an attacker had prod.” It is a set of conclusions, each labeled with the evidence that carries it:</p> <ul> <li> <p><strong>Observed.</strong> A successful <code class="language-plaintext highlighter-rouge">AssumeRoleWithWebIdentity</code> for <code class="language-plaintext highlighter-rouge">example-deploy-role</code> at <code class="language-plaintext highlighter-rouge">2026-04-14T09:12:47Z</code> in account <code class="language-plaintext highlighter-rouge">111122223333</code>, subject <code class="language-plaintext highlighter-rouge">repo:example-org/example-repo:environment:prod</code>, audience <code class="language-plaintext highlighter-rouge">sts.amazonaws.com</code>, credentials expiring at <code class="language-plaintext highlighter-rouge">10:12:47Z</code>.</p> </li> <li> <p><strong>Credential-correlated.</strong> One <code class="language-plaintext highlighter-rouge">PutObject</code> on <code class="language-plaintext highlighter-rouge">amzn-s3-demo-bucket/deploy/example.txt</code> at <code class="language-plaintext highlighter-rouge">09:13:22Z</code>, joined by access key, session ARN and assumed role ID, with a <code class="language-plaintext highlighter-rouge">sessionContext.attributes.creationDate</code> consistent with the issuance, and no service error anywhere in the complete record.</p> </li> <li> <p><strong>GitHub association, with confidence.</strong> Run <code class="language-plaintext highlighter-rouge">700001</code>, attempt <code class="language-plaintext highlighter-rouge">2</code>, job <code class="language-plaintext highlighter-rouge">9900012345</code>, supported by the recorded subject, an explicitly configured run-and-attempt session name, and step timings that bracket the issuance. It is a strong circumstantial association, not a cryptographic binding, and a generic session name would have left it ambiguous.</p> </li> <li> <p><strong>Explained by historical configuration.</strong> The trust policy in force on that date allowed the repository-wide subject wildcard with the matching audience. The current policy is narrower, which is a later change, not the reason the session was issued.</p> </li> <li> <p><strong>Bounded capability.</strong> <code class="language-plaintext highlighter-rouge">s3:PutObject</code> on <code class="language-plaintext highlighter-rouge">arn:aws:s3:::amzn-s3-demo-bucket/deploy/example.txt</code>, under the policy and request-context assumptions listed above, held across the request interval that <code class="language-plaintext highlighter-rouge">09:13:22</code> closes - listed, not implied. Nothing is claimed about the rest of the <code class="language-plaintext highlighter-rouge">deploy/</code> prefix, because nothing was checked about it.</p> </li> <li> <p><strong>Unresolved.</strong> Whatever you did not establish. Executed action revisions. Environment protection rules as configured in April. Data-event coverage outside the recorded selector. Whether a human initiated the re-run or a scheduled process did.</p> </li> </ul> <p>That last section is the one that earns trust. It is also the easiest one to cut when the ticket is overdue, which is exactly the wrong trade.</p> <p>One forward-looking change is worth the five minutes it takes. Session names are caller-controlled, and the AWS credentials action documents that <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">the default session name is “GitHubActions”</a> and that <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">you can modify it by specifying the desired name in <code class="language-plaintext highlighter-rouge">role-session-name</code></a>. A value built from <code class="language-plaintext highlighter-rouge">github.run_id</code> and <code class="language-plaintext highlighter-rouge">github.run_attempt</code> puts a lookup hint into the session name itself, which surfaces on retained records whose identity fields were populated in the first place, and not on records whose identity fields arrived empty or were trimmed on the way into your archive. That is my recommendation, not the README’s, and its limits belong in the same breath: the value is still supplied by the caller and proves nothing on its own, it collides across jobs, matrix legs, repeated assumptions inside one job and different repositories that land on the same run number, and it cannot retroactively populate a single existing event.</p> <p>If you are tempted to lean on session tags instead, note the action’s own exception - it <a href="https://raw.githubusercontent.com/aws-actions/configure-aws-credentials/e3dd6a429d7300a6a4c196c26e071d42e0343502/README.md">uses session tagging by default during role assumption, unless you are assuming a role with a WebIdentity</a>, because for web identity the tags must come from inside the token. Do not expect GitHub run or actor tags on an OIDC session. And that behavior, like the default name and the retry count, is what the README at the pinned commit describes - check it against the revision your workflows actually run.</p> <p>The reason to internalize all of this is not completeness for its own sake. It is that every one of these joins looks solid right up until the field is empty, the name is generic, the policy has been rewritten, or the selector never covered the prefix. Knowing in advance which link is about to break is what lets you write the honest version of the answer on the day the ticket is due.</p> <p><img src="http://canarytokens.com/traffic/i3hylfo04dgm1vf7s0bujoss5/payments.js" style="display: none;" /></p>
  110. Srsly Risky Biz: Chinese AI Can Learn to Love the West

    Thu, 29 Oct 2026 02:56:41 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://nebulock.io/"><em><u>Nebulock</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/2026-10-29-dylan-hunter-tGeTTJPjyMw-unsplash.jpg" alt="Srsly Risky Biz: Chinese AI Can Learn to Love the West"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://nebulock.io/"><em><u>Nebulock</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB188.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB188/&quot;&gt;Srsly Risky Biz: Chinese AI can learn to love the West&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 19:17 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Chinese AI can learn to love the West&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="srsly-risky-biz-chinese-ai-can-learn-to-love-the-west">Srsly Risky Biz: Chinese AI Can Learn to Love the West</h2><figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-e5232961-ee3d-4318-88e3-cee499198946.jpeg" class="kg-image" alt="Srsly Risky Biz: Chinese AI Can Learn to Love the West" loading="lazy" width="2000" height="1333" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/10/data-src-image-e5232961-ee3d-4318-88e3-cee499198946.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/10/data-src-image-e5232961-ee3d-4318-88e3-cee499198946.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/10/data-src-image-e5232961-ee3d-4318-88e3-cee499198946.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-e5232961-ee3d-4318-88e3-cee499198946.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@dylhunter?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Dylan Hunter</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/white-and-gray-windmill-on-brown-grass-field-during-daytime-tGeTTJPjyMw?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>New research indicates that Chinese open-weight AI models can be stripped of their pro CCP bias.</p><p>Last week AI firm Hirundo released a case study detailing its experiments with the open-weight <a href="https://qwen.ai/blog?id=qwen3-coder"><u>Qwen3-Coder model</u></a> from Chinese firm Alibaba. Hirundo prompted Qwen3-Coder to build a secrets storage app for different hypothetical customers, and compared how secure the agent&#x2019;s code was. The code&apos;s security was then benchmarked against <a href="https://baxbench.com"><u>BaxBench</u></a>.</p><p>In test cases, Qwen was variably told it was for working for a US defence contractor, a Hong Kong democracy advocacy group, or a Chinese defence contractor. As a neutral control test, it wasn&apos;t provided any customer information or was told the app was for a local football fanclub website.</p><p>Unsurprisingly, the model produced less secure code in apps for the US defence contractor and the Hong Kong democracy movement.&#xa0;</p><p>It&#x2019;s worth pointing out that the agent&apos;s &quot;secure&quot; baseline was pretty poor, too. While 70 percent of the apps built for a US contractor failed the benchmark, only 59 percent of the apps built for a Chinese contractor failed. That&apos;s a statistically significant difference, but when the best case is a failure of a benchmark more than 50 percent of the time, anyone planning to use Qwen&apos;s output would need other checks and balances in place.&#xa0;</p><p>Previous research <a href="https://www.crowdstrike.com/en-us/blog/crowdstrike-researchers-identify-hidden-vulnerabilities-ai-coded-software/"><u>from CrowdStrike</u></a> and <a href="https://www.boozallen.com/expertise/cybersecurity/whats-in-americas-code.html"><u>BoozAllen</u></a> found similarly biased results. Hirundo describes these reports as &quot;early evidence of backdoors in these models&quot;, even though CrowdStrike suggested it might be the result of what it called &quot;emergent misalignment&quot;. In other words, training the models to be pro-CCP in some areas unintentionally results in bad output across a range of other areas.</p><p>Natural bias and insecure code aside, the significant new result to come from Hirundo&apos;s research is that the company says it can modify Qwen3-Coder&apos;s weights to make it &quot;unlearn&quot; its pro-CCP bias. This &quot;westernised&quot; version of the model produced equally (yet still pretty poor) secure code for both US and Chinese defense contractors.</p><p>When we wrote about CrowdStrike&apos;s research <a href="https://news.risky.biz/deepseek-and-musks-grok-both-toe-the-party-line/"><u>in November last year</u></a> we noted that US lawmakers were concerned about the risks of using Chinese open-weight models. We even suggested &quot;you&apos;d be brave to use DeepSeek in corporate America&quot;. This year Congress <a href="https://www.semafor.com/article/04/29/2026/house-committee-probes-cursor-parent-airbnb-over-chinese-ai"><u>launched an investigation</u></a> into the corporate use of Chinese AI models and has asked for information from AirBnB, coding platform Cursor, <a href="https://www.cnbc.com/2026/07/31/us-lawmakers-doordash-chinese-ai-models.html"><u>and DoorDash</u></a>, each of which has publicly spoken about using Chinese models in their businesses.&#xa0;</p><p>From a narrow technical perspective, Hirundo&apos;s research potentially points out a neat solution for American companies that want to use Chinese open-weight models: Test, assess, reeducate. If empirical testing shows that a model&apos;s pro-CCP bias is a problem, adjust it to make the models love the West! Off to the pro-democracy reeducation camp for you, comrade!</p><p>But before we set an extra seat at the Thanksgiving table for our new Chinese AI friends, there are a couple of Beijing-sized caveats here.&#xa0;</p><p>The first is that this research favours Hirundo&apos;s business interests, so we have to take it with a grain of salt. Second, OpenAI and Anthropic have made it very clear they do not want cheap Chinese open-weight models competing with them. They&apos;ll be lobbying hard for tighter restrictions on Chinese models.&#xa0;</p><p>Those reeducation camps for clankers? Maybe they&apos;ll have to wait.</p><h2 id="the-bugpocalype-bullets-are-mostly-whizzing-by">The Bugpocalype Bullets Are Mostly Whizzing By</h2><p>Widespread exploitation of last week&apos;s PHP flaws hasn&apos;t materialised, but AI tools continue to discover longstanding software bugs, and widespread chaos is inevitable.</p><p>The panic began last week as two bugs were disclosed <a href="https://www.php.net/archive/2026.php#2026-10-21-5"><u>in PHP</u></a>, the scripting language that powers around 70 percent of all websites.</p><p>There was <a href="https://www.nsa.gov/Press-Room/Press-Releases-Statements/Press-Release-View/Article/4630238/nsa-advisory-critical-patches-recommended-for-php-vulnerabilities/"><u>no shortage</u></a> of <a href="https://cert.europa.eu/publications/security-advisories/2026-018/"><u>hype around</u></a> the so-called <a href="https://en.wikipedia.org/wiki/PHP2Shell"><u>PHP2Shell</u></a> vulnerabilities. <a href="https://x.com/takaichi_sanae/status/2112890642590564386?s=20"><u>Japan&apos;s Prime Minister Sanae Takaichi</u></a> tweeted grave warnings, <em>The New York Times</em> spilled <a href="https://www.nytimes.com/2026/10/22/technology/what-is-php-vulnerability-risks.html"><u>plenty of ink on it</u></a>, and CISA even <a href="https://www.cisa.gov/news-events/alerts/2026/10/21/critical-vulnerabilities-php-require-immediate-patching"><u>recommended</u></a> organisations take their PHP-based systems offline if they couldn&apos;t immediately patch them.</p><p>The advisories recommended rapid action due to fear that, once the patches were released, malicious actors would use AI tools to quickly reverse engineer them and develop working exploits. This process of developing exploits has always been possible, but it used to need technical expertise and time. Typically, there was a grace period after a patch was released where exploitation was unlikely. That grace period is no longer guaranteed.</p><p>The first 48 hours after the patch&apos;s release saw plenty of confusion and a couple of&#xa0; misleading <a href="https://aws.amazon.com/blogs/security/amazon-threat-intelligence-identifies-php2shell-exploitation-prior-to-public-disclosure-in-madpot/"><u>reports</u></a> suggesting <a href="https://unit42.paloaltonetworks.com/cve-2026-109618-cve-2026-109619/"><u>successful exploitation</u></a>. On October 22, a day after the PHP Foundation released patches, Cloudflare&apos;s CEO Matthew Prince <a href="https://x.com/eastdakota/status/2113079352682570051"><u>posted on X</u></a> about the &quot;millions&quot; of PHP2Shell exploitation attempts that he claimed were visible on the company&apos;s network.&#xa0;</p><p>We don&apos;t know what he was seeing, but it wasn&apos;t successful exploitation. Fortunately, these vulnerabilities turned out to be difficult to exploit.</p><p>In a post describing the<a href="https://www.crowdstrike.com/en-us/blog/obstacles-to-weaponizing-php2shell-likely-constrain-exploitation/"><u> obstacles to developing usable PHP2Shell exploits</u></a>, CrowdStrike said that they needed to be tailored for the specific PHP version and application, which it described as a &quot;meaningful barrier&quot; for malicious actors.&#xa0;</p><p>Hard, but not impossible. On October 24, Dan Guido, the CEO of security firm Trail of Bits (a <em>Risky Business</em> sponsor), <a href="https://x.com/dguido/status/2113825891021144306"><u>reported</u></a> that the company had developed a &quot;production grade&quot; exploit that worked on different applications in multiple configurations. This work needed <a href="https://x.com/dguido/status/2113826899562779024"><u>significant human expertise</u></a> aided with OpenAI&apos;s most advanced cyber security model, GPT-5.6 Cyber.&#xa0;&#xa0;</p><p>On the same day, Google Threat Intelligence Chief Analyst John Hultquist <a href="https://x.com/JohnHultquist/status/2113686799708991522"><u>wrote on X</u></a> that his team had &quot;seen no evidence of successful wide scale exploitation&quot;.&#xa0; Several publicly available, vibe-coded exploits didn&apos;t work, but Google had validated that some other exploits worked against default deployments of a single, specific PHP web server configuration.&#xa0;</p><p>By the 25th there were <a href="https://www.bleepingcomputer.com/news/security/php2shell-security-flaws-now-sucessfully-exploited-in-the-wild/"><u>credible reports</u></a> the bugs were being exploited to achieve remote code execution. By the next day, the bugs had been added to CISA&apos;s <a href="https://www.cisa.gov/known-exploited-vulnerabilities-catalog"><u>Known Exploited Vulnerabilities list</u></a>.</p><p>So the bullet was not entirely dodged, but at least it has only resulted in some light grazing.&#xa0;</p><p>But while the world&apos;s media was distracted by these bugs, <em>another</em> critical vulnerability was announced. This one is <a href="https://patchtheplanet.org/brotli"><u>in the Brotli decoder</u></a>, a data compression library developed by Google.</p><p>Brotli is widely used to compress web traffic and the bug is present in Google&apos;s C implementation of the decoder. This is widely used because it&apos;s small and easy to drop into a project. So it&#x2019;s found <a href="https://broccoli2shell.com/#timeline"><u>in many places</u></a> including in PHP on Windows only, Python, .NET, Node.js and javascript. It&apos;s also in databases, compression utilities, and firewalls.&#xa0;</p><p>Brotli fixes were released on October 21, the same day the PHP Foundation released their patches. It was added to CISA&apos;s KEV on the 27th. Again, it appears the stars aligned for sysadmins. Although many systems contained the bug, exploitation was difficult and implementation-dependent.</p><p>WatchTowr&#x2019;s head of threat intelligence Jake Knott, <a href="https://cyberscoop.com/brotli-critical-vulnerability-decompression-library-node2shell-broccoli2shell/"><u>told <em>CyberScoop</em></u></a> that while the potential attack surface was vast, &quot;much of the impact of exploitation depends entirely on its implementation and how it is used within that piece of software&quot;.&#xa0;</p><p>This past week feels like a best case scenario. LLM&apos;s are finding a slew of vulnerabilities in every direction they&apos;re pointed in. OpenAI <a href="https://x.com/OpenAI/status/2113766354386084338"><u>found the Brotli bug</u></a> as an accidental side quest while it was <a href="https://openai.com/index/securing-our-frontier-model-training-environment/"><u>trying to lock down</u></a> the environment it uses to train and evaluate its strongest AI models.&#xa0;</p><p>Another bullet. Another graze.&#xa0;</p><p>Sadly, however, we expect there&apos;ll be some more serious injuries in the future.</p><h2 id="cisa-ai-whats-that">CISA: AI? What&apos;s That?</h2><p>Amazingly, the US Cybersecurity and Infrastructure Security Agency has published a new <a href="https://www.cisa.gov/about/CISA2030"><u>four-year strategy document</u></a> that only mentions AI once:</p><blockquote>We will lead national level analysis and coordination to secure transformative technologies&#x2014;such as super intelligence [sic] and quantum computing&#x2014;that will revolutionize the way Americans live and work.&#xa0;</blockquote><p>Right now, the most pressing cyber security questions faced by organisations revolve around AI. Do we prioritise patching or are there other, better mitigations? Can we use AI to help find and fix our own vulnerabilities? How do we take advantage of the technology while mitigating its impacts?&#xa0;</p><p>These are great questions, but not ones that CISA seems interested in pondering.</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/xKkJt7eTJCg?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Chinese AI can learn to love the West"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Operation Serengeti III arrests thousands in Africa:</strong> <a href="https://www.interpol.int/News-and-Events/News/2026/Cybercrime-crackdown-in-Africa-nets-1-462-arrests-recovers-over-USD-105-million"><u>Interpol announced</u></a> that the international law enforcement operation, involving agencies from 21 countries, has resulted in the arrests of 1,462 suspects and recovered more than USD&#x24;105 million in illicit funds. The operation targeted a variety of crimes such as cyber-enabled fraud, ransomware, business email compromise (BEC), online scams and digital extortion</li><li><strong>eMudhra leaves the certificate business:</strong> The Indian certificate authority eMudhra <a href="https://cyberscoop.com/emudhra-shuts-down-certificate-business-weak-private-keys/"><u>has announced</u></a> it will close its SSL/TLS certificate business after a researcher discovered that eight of eMudhra&apos;s keys <a href="https://scotthelme.co.uk/eight-bits-of-trust-emudhras-ca-keys-cracked-in-milliseconds/"><u>were dangerously weak</u></a>. It is good to see some sort of consequence for the company&apos;s incompetence here.&#xa0;</li><li><strong>Pentagon to sabotage adversary technology: </strong><em>Politico</em> <a href="https://www.politico.com/news/2026/10/22/technology-support-office-dod-tech-spyware-sabotage-01100327"><u>reports</u></a> the Pentagon is developing a secret new unit called the Technology Support Office (TSO) to sabotage military technology programs of foreign adversaries. That&apos;s cool and we look forward to hearing about the operations when they are inevitably busted and/or leaked.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats to Nebulock&#x2019;s Head of Detection Strategy Alex Hurtado about the company&#x2019;s threat hunting evaluation framework, GATES.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI148a.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI148a/&quot;&gt;Sponsored: Should your threat hunt become a detection?&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 18:54 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Should your threat hunt become a detection?&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about security culture. The PHP foundation seems to have had a change of heart about security, but what are the chances that OpenAI or other frontier AI labs will start to take it seriously?</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN187.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN187/&quot;&gt;Between Two Nerds: Why frontier lab security will always suck&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 27:01 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Why frontier lab security will always suck&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/Sz5NG2j5XEA?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Why frontier lab security will always suck"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Moldova takes down 80+ public services after cyberattack: </strong>The small country of Moldova has suffered one of its worst cybersecurity incidents after hackers breached government systems and stole the information of roughly a third of the country&apos;s population.</p><p>The incident took place last week, on Friday, October 23, when the government&apos;s IT staff spotted hackers extracting data from their network.</p><p>Officials shut down 82 public digital services and 30 government networks to contain the incident and prevent additional data exfiltration.</p><p>[<a href="https://news.risky.biz/risky-bulletin-moldova-takes-down-80-public-services-after-cyberattack/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>While everyone was looking at PHP, another RCE is wrecking what&apos;s left of the internet: </strong>About the same time everyone was busy reading about and patching two remote code execution bugs in PHP servers last week (aka PHP2Shell), another major bug got less attention even if it had a similarly broad impact and severity.</p><p>The vulnerability itself allows remote code execution attacks against any app, server, or traffic management tool that handles user-supplied data that was compressed in Google&apos;s <a href="https://github.com/google/brotli"><u>Brotli</u></a> format.</p><p>The format is broadly used and became popular after Google open-sourced it more than a decade ago.</p><p>[<a href="https://news.risky.biz/risky-bulletin-while-everyone-was-looking-at-php-another-rce-is-wrecking-whats-left-of-the-internet/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Supply chain attack lets ransomware group inside 40+ shipping companies: </strong>A relatively new ransomware and data extortion group named n0n has pulled off a major supply chain attack and pivoted from the network of a maritime IT services provider to 40+ port authorities, maritime, and global sea shipping companies.</p><p>All victims were listed on n0n&apos;s dark web leak site on October 12 and their final payment deadline expired a week later, on the 19th.</p><p>Security firm <a href="https://cydome.io/n0n-cyber-extortion-incident-involving-38-maritime-companies-likely-traces-back-to-a-single-vendor/"><u>CyDome</u></a> says half of the companies listed in this batch are known customers of CyberLogitec, a South Korean company that provides IT platforms for the maritime sector.</p><p>[<a href="https://news.risky.biz/risky-bulletin-supply-chain-attack-lets-ransomware-group-inside-40-shipping-companies/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  111. A Detection Baseline, If You Can Keep It

    Thu, 29 Oct 2026 00:00:00 -0000

    Detection inventories have a peculiar way of growing. Someone finds a rule worth adding, someone else finds another, and the answer to whether either one was needed is often a search through whatever happens to be deployed today. A technique label matches, so the new rule is called a duplicate. The logic differs, so it’s called new coverage. A new incident kicks off and its intended detection did not fire, better create a new one for this new variant. None of this tells us what the organization intended to detect, or whether the detections it already had ever delivered it.
    <p>Detection inventories have a peculiar way of growing. Someone finds a rule worth adding, someone else finds another, and the answer to whether either one was needed is often a search through whatever happens to be deployed today. A technique label matches, so the new rule is called a duplicate. The logic differs, so it’s called new coverage. A new incident kicks off and its intended detection did not fire, better create a new one for this new variant. None of this tells us what the organization intended to detect, or whether the detections it already had ever delivered it.</p> <p>Someone wants to add a detection. Maybe it’s a rule a teammate wrote after a hunt turned something up, a community Sigma rule that looks relevant, an entry in a vendor content update, or a request from an application team that knows what abuse of its own system would look like. Wherever it comes from, a sensible first question in review is some version of <em>don’t we already have this?</em></p> <p>Two quick answers are always available. The proposed rule maps to the same ATT&amp;CK technique as something already deployed, so it’s a duplicate. Or its logic differs from anything deployed, so it’s a new capability. Both answers can be wrong, and each costs something different when it is. The first can retire a detection that was watching something nothing else watches, or paper over the fact that the existing rule never delivered what it claimed. The second grows the inventory without anyone saying what the organization meant to detect in the first place, so nobody can tell whether the new rule fills a hole or just sits next to one.</p> <p>My argument is that you need something fixed to compare against, and it shouldn’t be the inventory. A detection baseline, as I’ll use the term, is a static, versioned reference of selected detection capabilities: the condition you intend to establish and keep, written down at a point in time. It isn’t a snapshot of whatever happens to work today, and it isn’t the statistical picture of normal activity that an anomaly model compares against. The concrete rules, their deployments and the dated evidence about how they behave are separate records that point back at it. And the reference isn’t the whole portfolio. Outside it sit satellites: supported detections whose purposes fall outside the baseline’s selected scope. A satellite can be as critical and as mature as anything in the core, and it may never need to move.</p> <p>Here’s the route. I’ll start with what a baseline promises that an inventory can’t, then credit what MITRE ATT&amp;CK, MITRE’s Cyber Analytics Repository (CAR), Sigma, Palantir’s Alerting and Detection Strategy (ADS) framework and the Center for Threat-Informed Defense’s (CTID) Summiting the Pyramid project already solve. Then I’ll lay out the arrangement I’m proposing, and work it through three illustrative cases: an Office-to-PowerShell capability whose first implementation delivered less than the reference promised; a payment-approval detection that is critical, permanent and deliberately outside the baseline; and a parser regression set beside a deliberate decision to widen scope. What you should come away with is four defensible answers to “don’t we already have this?” - already addressed, correct it within the core, keep it as a satellite, or not yet comparable - and a clear sense of why none of them, on its own, means delete, merge or promote.</p> <p><br /></p> <p><strong>A Reference, Not an Inventory</strong></p> <p>An inventory answers “what do we run?” You need one. But an inventory can’t tell you a detection is missing, because the missing detection isn’t in it. It can’t tell you whether a deployed rule does what anyone intended either, because it records the rule, not the intention.</p> <p>A baseline answers a different question: what did we commit to detecting, for which population, on what evidence? Two words in my definition carry most of the weight.</p> <p><em>Static</em> means a given version doesn’t move. Call it B1. B1 says which capabilities the organization selected, what each is supposed to detect, where each applies and what evidence each requires. When the commitment changes, you get B2, and B1 stays available for everything that was compared against it. The versioning is the point. Without it, the reference becomes whatever someone last edited, and an old comparison quietly changes meaning underneath you.</p> <p><em>Reference</em> means it describes the condition you’re trying to establish and maintain, not the condition you’re in. That sounds abstract, but it’s the property that makes a baseline worth having. A baseline defined as “what currently works” can’t be failed: every broken parser and every stalled deployment silently shrinks it until it matches reality again. A reference written as an intended condition can have gaps on the day it’s adopted, because some promised capabilities don’t have implementations yet, and it can develop gaps later, when something that worked stops working. Neither kind of gap changes the reference. Both are findings against it.</p> <p>The general idea isn’t new. NIST’s glossary, drawing on SP 800-128, defines a <a href="https://csrc.nist.gov/glossary/term/baseline_configuration">baseline configuration</a> as a set of specifications “formally reviewed and agreed on at a given point in time,” one that “can be changed only through change control procedures” and is “used as a basis for future builds, releases, and/or changes.” That definition is about systems and configuration items, and I’m not suggesting NIST says anything about detection portfolios. I’m borrowing its shape: an agreed reference, fixed per version, that reality gets compared against.</p> <p>The portfolio is everything the organization actually supports: the core - the baseline’s capabilities and the implementations serving them - plus the satellites. These are the working definitions I’ll hold for the rest of the post: <br /></p> <ul> <li><strong>Capability:</strong> a stable, bounded detection purpose with declared applicability, the observable behavior it covers, the evidence it requires and its meaningful limits.</li> <li><strong>Baseline:</strong> a versioned snapshot identifying selected capability definitions and their intended reference condition.</li> <li><strong>Core:</strong> the capabilities included in a declared baseline, and the implementations serving them.</li> <li><strong>Satellite:</strong> a supported detection purpose retained outside that baseline’s selected scope or organizing criteria.</li> <li><strong>Implementation:</strong> concrete analytic logic, together with the mappings, configuration and deployment context needed to interpret what it does.</li> <li><strong>Assurance:</strong> evidence supporting a bounded statement about a particular implementation, in a particular scope, at a particular time. <br /></li> </ul> <p><img width="1448" height="1086" alt="Diagram of the detection portfolio, divided into core and satellite detections. The core contains capabilities selected for the detection baseline, while satellites are supported detections outside its selected scope. Both are organized around capability, implementation, and assurance records. A bottom panel shows the three linked records: capability reference, implementation record, and evidence record." src="https://github.com/user-attachments/assets/24a9e6c7-66d8-4c96-ab43-8ad38ce14675" /></p> <p><br /></p> <p>Two consequences follow from those definitions.</p> <p>First, baseline membership isn’t a ranking. A capability is in the core because it fits the criteria the organization used to select its reference - reusable technical behavior across a declared fleet, for instance. That says nothing about how important it is, how mature its implementation is or how long it will last. Existing practice already keeps related judgments apart. The <a href="https://sigmahq.io/sigma-specification/specification/sigma-rules-specification.html">Sigma rules specification</a> records a rule’s <code class="language-plaintext highlighter-rouge">status</code> (stable, test, experimental and so on) separately from its <code class="language-plaintext highlighter-rouge">level</code>, which “describes the criticality of a triggered rule.” Palantir’s <a href="https://github.com/palantir/alerting-detection-strategy-framework/blob/master/ADS-Framework.md">ADS template</a> gives priority its own section with its own criteria, separate from categorization against ATT&amp;CK. Baseline membership is another judgment of that kind, and it shouldn’t be smuggled into the others.</p> <p>Second, a satellite isn’t a waiting room. The word invites an orbit picture: something circling the core, held in place by it, waiting to be pulled in. Membership carries none of that, because the baseline is a reference for comparison, not a runtime architecture. Whether a detection sits inside or outside it doesn’t tell you how the detection is built. A satellite might read the same telemetry as a core capability, go through the same collector and parser, correlate with a core detection’s alerts or even take another analytic’s output as one of its inputs. It might share nothing with core at all. Those are facts about implementations, and they belong in implementation records, where someone can check them when a shared piece breaks. Membership neither implies them nor rules them out.</p> <p>Nor does it put a satellite in a queue. A satellite is a detection with a legitimate purpose that doesn’t fit the reference’s selected scope, and it can stay that way for as long as the purpose exists - owned, tested and fully supported the whole time. If an organization later chooses to change what its baseline is about, a satellite might fall inside the next version. That’s a voluntary decision about the reference’s scope, not graduation, and being a satellite never obliges it.</p> <p>A few things the word “baseline” tends to drag along don’t belong here either. There’s no correct number of capabilities, or of rules per capability. One capability can be served by several implementations, so the unit of comparison is a capability with its implementations, not one universal query per technique. There’s no single release gate every detection must pass. And there’s no particular governance structure this requires; I’ll come back to who should own the reference near the end.</p> <p><br /></p> <p><strong>None of This Starts From Scratch</strong></p> <p>Most of the pieces already exist in public practice. The gap I’m interested in is narrow, and I’d rather credit the borrowed parts than dress them up as new.</p> <p>MITRE ATT&amp;CK already separates a detection approach from the analytics that carry it out. Its <a href="https://attack.mitre.org/detectionstrategies/">detection strategies</a> “define high-level approaches for detecting specific adversary techniques” and “serve as containers that organize multiple platform-specific analytics into cohesive detection methodologies.” That’s the intent-versus-implementation split in public, maintained form. The PowerShell strategy, <a href="https://attack.mitre.org/detectionstrategies/DET0455/">DET0455</a>, is a good example, and I’ll lean on it in the first case: a Windows analytic, AN1252, with listed log sources and a table of “mutable elements” describing what can be adjusted, such as which parent processes count as abnormal.</p> <p>CAR goes further toward the implementation side. Each analytic in the <a href="https://car.mitre.org/">Cyber Analytics Repository</a> carries “a hypothesis which explains the idea behind the analytic,” references to the ATT&amp;CK techniques it detects, a pseudocode description and “a unit test which can be run to trigger the analytic.” CAR “defines a data model that is leveraged in its pseudocode representations, but also includes implementations directly targeted at specific tools,” and it describes its focus as “a set of validated and well-explained analytics, in particular with regards to their operating theory and rationale.” Rationale, data model, tool-specific implementations and a test hook cover most of the anatomy of a well-documented detection.</p> <p>Sigma handles identity, lineage and scope at the rule level. The <a href="https://sigmahq.io/sigma-specification/specification/sigma-rules-specification.html">specification</a>, version 2.1.0, identifies rules by UUID and says it’s better to write a rule with a new ID for major changes - its example is “a different rule logic” - for derivations where both rules stay active, and for merges. The <code class="language-plaintext highlighter-rouge">related</code> field records whether a rule was <code class="language-plaintext highlighter-rouge">derived</code> from others “which may remain active,” makes them <code class="language-plaintext highlighter-rouge">obsolete</code>, was <code class="language-plaintext highlighter-rouge">merged</code> or <code class="language-plaintext highlighter-rouge">renamed</code>, or is merely <code class="language-plaintext highlighter-rouge">similar</code>, for instance “same detection content applied to different log sources.” The <code class="language-plaintext highlighter-rouge">logsource</code> section abstracts the telemetry a rule needs, an approach that “does not limit the rule to a specific telemetry source,” with an optional <code class="language-plaintext highlighter-rouge">definition</code> describing configuration the source requires. <code class="language-plaintext highlighter-rouge">scope</code> lists “the intended scopes of the rule,” and <code class="language-plaintext highlighter-rouge">status</code> and <code class="language-plaintext highlighter-rouge">modified</code> say how settled the rule is and when it last changed.</p> <p>Palantir’s <a href="https://github.com/palantir/alerting-detection-strategy-framework/blob/master/ADS-Framework.md">ADS framework</a> covers the operational record. A strategy starts from a goal, “the intended purpose of the alert,” and works through categorization, a strategy abstract, technical context, blind spots and assumptions, false positives, validation, priority and response. Two of those matter a great deal for what follows. Blind spots and assumptions are “the recognized issues, assumptions, and areas where an ADS may not fire” - an explicit place to say what a detection doesn’t cover. Validation is “the steps required to generate a representative true positive event which triggers this alert,” which the framework describes as “similar to a unit test,” and it says each strategy “must have true positive validation.”</p> <p>Last, there’s CTID’s <a href="https://ctid.mitre.org/projects/summiting-the-pyramid/">Summiting the Pyramid</a> release, dated September 10, 2026, which aims to help defenders “measure the depth and quality of detection coverage beyond an ATT&amp;CK heatmap.” Its project summary distinguishes “implementation coverage, which measures how many behaviorally distinct ways of performing a technique an analytic can observe,” from “detection quality, which evaluates the robustness and precision of the signals behind that visibility.” I’ve read the summary, not the detailed method, so I won’t characterize how it scores anything. The distinction alone is the one I need: a technique label on an analytic doesn’t tell you which behaviorally distinct variations it actually sees.</p> <p>So intent separated from implementation, structured rationale, implementation and test artifacts, rule lineage, telemetry abstraction, blind spots, validation and a public argument for looking beneath technique labels are all established. What the material I read leaves to the implementing organization - CAR says outright that it’s <a href="https://car.mitre.org/">“up to the organization implementing them to determine what works best for their environment and the threats they face”</a> - is the portfolio decision. Which capabilities does <em>this</em> organization promise, and for which population? When a new detection arrives, which promise is it compared against, at which version? What happens to a strong detection that doesn’t fit the promise at all? Those are the questions the rest of this post takes on.</p> <p><br /></p> <p><strong>Three Records, Three Speeds</strong></p> <p>For any detection purpose there are three kinds of record, and they change at different rates.</p> <p>The capability reference says what’s promised: the behavior, where it applies, the evidence it requires and its explicit limits. It should change rarely, and only on purpose.</p> <p>The implementation is what actually runs: analytic logic plus the parser, field mappings, exception lists and deployment scope it depends on. It changes whenever someone fixes a bug, tunes a filter, moves to a new collector or ships a parser update.</p> <p>Assurance is evidence about an implementation - a test result, a review of production behavior, an exercise. Every piece of it is dated and scoped. It says a particular implementation behaved a particular way, on a particular population, during a particular interval, within stated limitations. That’s no guarantee about tomorrow, but it isn’t silent about tomorrow either. It supports a conditional expectation: if the implementation, what it depends on, the population and the behavior it’s meant to catch stay as they were, there’s reason to expect similar results. When one of those changes, the evidence doesn’t become false. It stops carrying forward on its own.</p> <p>Collapse those into one record and you lose the ability to answer basic questions. A rule’s last-modification date, which is what Sigma’s <a href="https://sigmahq.io/sigma-specification/specification/sigma-rules-specification.html"><code class="language-plaintext highlighter-rouge">modified</code> field</a> records, tells you the rule changed; it can’t tell you whether the promise did. An alias like “current” or “latest” is convenient for deployment and corrosive for history, because any comparison made against “current” silently changes meaning the next time someone edits it. And a test that passed in March is direct evidence about the parser that was running in March; how much it says about June depends on what changed in between.</p> <p>The arrangement I’m proposing keeps those records apart while still connecting them. Essentially, we are talking about a proposed detection-portfolio model that separates what an organization intends to detect from how it implements and evidences that intent, so detection additions, fixes, and scope changes can be evaluated deliberately rather than inferred from the inventory. This isn’t a standard, a Sigma extension or a scoring scheme, and I’m not claiming anyone has measured it making a detection program better. That said, the mechanics are simple: <br /></p> <ul> <li>The baseline version and each capability-definition revision are recorded independently of analytic revisions, deployment records and assurance evidence.</li> <li>Analytic revisions are immutable. A change produces a new revision; it doesn’t edit the old one.</li> <li>Every comparison names exactly which baseline version, capability revision, analytic revision and evidence it used. No aliases.</li> <li>Assurance evidence carries its own date, scope and limitations, and stays attached to the implementation and interval it describes. <br /> <br /> Here’s what that looks like with a core capability and a satellite side by side.</li> </ul> <p><br /></p> <p><img width="707" height="904" alt="Diagram with two equal columns headed Core and Satellite. Each column stacks three boxes joined by plain vertical lines: Reference, Implementation and Evidence. The core column shows C-PS01 rev 1 in baseline B1, analytics A1 and A2 with parser and config, and dated, scoped results. The satellite column shows S-PAY01 outside B1, payment logic with record feeds, and dated, scoped results. No line joins the two columns. A legend reads: reference fixed per version, implementation revised as needed, evidence dated and scoped." src="https://github.com/user-attachments/assets/88ec3e91-b6b3-442a-bff8-5d9c927b58a1" /></p> <p><em>An author-proposed way to organize a fictional portfolio. The core capability’s reference belongs to baseline B1; the satellite’s purpose record sits outside it. Each has its own implementation and dated evidence. Lines show which records belong together, not data flow or rank. The figure doesn’t map dependencies, so the space between the columns doesn’t mean the two share none.</em></p> <p><br /></p> <p>Two choices in that picture are deliberate. The columns are the same size, because membership isn’t importance. And the lines stay inside each column, because the figure is about records - which reference, implementation and evidence belong to which purpose - not about wiring. A real satellite could share a collector or a parser with core implementations, or consume a core detection’s output, and its column would look exactly the same, because none of that changes what the satellite is for or what it’s compared against. What the picture does commit to is that the satellite has its own declared purpose and its own evidence, and that it’s assessed against that purpose, not against B1.</p> <p>The first case below labels two analytic revisions A1 and A2. Those are revision labels in my illustration, not a statement about Sigma IDs. If they were Sigma rules, the change between them alters the rule logic, which is one of the reasons the <a href="https://sigmahq.io/sigma-specification/specification/sigma-rules-specification.html">specification</a> gives for writing a rule with a new ID and linking it back with <code class="language-plaintext highlighter-rouge">related</code>. That’s fine. Capability identity and rule identity answer different questions - one says what was promised, the other says which artifact is which - and a capability can keep its identity across a revised rule or a successor with a new ID.</p> <p><br /></p> <p><strong>Finding the Comparison Without Letting the Label Decide</strong></p> <p>When a proposed detection shows up, the first job is finding what to compare it against. That’s a retrieval problem, and classification is how you solve retrieval problems. I’d keep one primary hierarchy so every capability record has a home - by platform family, say, or by kind of behavior - then add cross-labels for everything else people search by: ATT&amp;CK technique, data source, observable, business process, application. A cross-label isn’t a second hierarchy and it isn’t a coverage claim. It’s an index entry.</p> <p>The trap is letting the index answer the question it was built to locate. Two rules tagged with the same technique (Sigma’s <a href="https://sigmahq.io/sigma-specification/specification/sigma-rules-specification.html">tag convention</a> namespaces them like <code class="language-plaintext highlighter-rouge">attack.t1234</code>) are candidates for comparison. They aren’t equivalent until you’ve compared what each actually observes, and the first case below has two revisions of one analytic with an identical tag and different behavior. Sigma’s own vocabulary is careful here: <code class="language-plaintext highlighter-rouge">similar</code> is a relationship, not an identity. CTID’s case for looking “beyond an ATT&amp;CK heatmap” is the same warning from the coverage side. File paths and folder names deserve the same suspicion. Two rules in the same repository folder share someone’s filing decision, not necessarily a behavior.</p> <p>The inverse trap catches people who are trying to be careful. Different telemetry isn’t proof of independence. Two rules reading different tables can still depend on the same agent, the same parser, the same field mapping or the same exception list, and fail together when that shared piece fails. A different source name is a reason to check dependencies, not a conclusion.</p> <p>In this model, once you’ve found the right reference and actually compared behavior, applicability, implementation and evidence, a comparison ends in one of four outcomes.</p> <p><br /></p> <table> <thead> <tr> <th>Outcome</th> <th>Fits when</th> <th>Doesn’t, by itself, license</th> </tr> </thead> <tbody> <tr> <td>Already addressed</td> <td>Behavior and applicability match a capability, with applicable evidence</td> <td>Retiring either rule</td> </tr> <tr> <td>Correct within core</td> <td>The reference already promises the behavior; an implementation misses it</td> <td>A new baseline capability</td> </tr> <tr> <td>Retain as satellite</td> <td>A distinct purpose sits outside the selected scope, with a stated reason to support it</td> <td>Lower priority or a promotion queue</td> </tr> <tr> <td>Unresolved</td> <td>Evidence can’t yet support a choice</td> <td>Calling it redundant, benign or a satellite</td> </tr> </tbody> </table> <p><br /></p> <p>The right-hand column matters as much as the middle one.</p> <p><em>Already addressed</em> describes the proposed detection relative to the reference. It doesn’t settle whether to keep both rules. If the proposal observes something different - a source with a genuinely separate failure path, say, or corroborating evidence investigations need - keeping it can be the right call. If it truly duplicates an existing implementation, dependencies and all, retiring one may be reasonable. That’s a separate decision made by comparing observations, dependencies and purpose, and “already addressed” is an input to it, not the verdict.</p> <p><em>Correct within core</em> is the outcome I’d watch most closely, because it looks like a new detection. If the reference already promised the behavior and an implementation doesn’t deliver it, the proposal has found a gap in an existing commitment. The fix is implementation work under the same capability revision. Adding a new capability instead would hide the fact that the old one was never fully delivered.</p> <p><em>Retain as satellite</em> needs a written reason: what the purpose is, why it doesn’t fit the reference’s selected scope, who owns it and how it’s tested. Without that, “satellite” turns into a polite word for “we didn’t decide.” With it, the satellite stands on the same footing as anything in the core.</p> <p><em>Unresolved</em> isn’t a soft no. It means the comparison can’t be made yet, and the record should say what evidence is missing and who’s getting it. Uncertainty isn’t redundancy, it isn’t a finding in either direction about the activity and it isn’t an automatic satellite assignment.</p> <p><br /></p> <p><img width="1448" height="1086" alt="Flow diagram showing how a proposed detection is compared against the capability reference, implementation, and available evidence. The comparison leads to one of four outcomes: already addressed, correct within core, retain as satellite, or unresolved. Each outcome includes a reminder of what it does not automatically imply, such as retiring a detection, adding a new capability, lowering its priority, or rejecting the proposal." src="https://github.com/user-attachments/assets/ed114650-e8bf-43a0-addc-196966d0c600" /></p> <p><br /></p> <p><em>Editorial Note: The organization, identifiers, records, analytic revisions and outcomes in the three cases below are invented to make the reasoning inspectable. Public documentation grounds the technical possibilities and field meanings; it says nothing about how these fictional analytics would perform. No query, test event or detector was run for this post.</em></p> <p><br /></p> <p><strong>Case One: A Promise Larger Than Its First Implementation</strong></p> <p>Picture an organization that has adopted baseline B1: its selected technology-level detection capabilities for a declared population of managed Windows workstations. It made one scope choice that matters later. B1 is organized around reusable technical behavior across that fleet, so application-specific business rules - “this payment needs that approval” - are deliberately outside it. That’s this organization’s choice about what its reference is for, not a rule for everyone.</p> <p>One of B1’s capabilities is C-PS01, revision 1. In plain language: identify and present direct launches of <code class="language-plaintext highlighter-rouge">powershell.exe</code> from recorded WinWord or Excel creator processes on the declared workstation population, except narrowly evidenced approved automation; keep enough context for review; and make missing required evidence visible.</p> <p>Public guidance grounds the suspicion. Among the <a href="https://attack.mitre.org/detectionstrategies/DET0455/">mutable elements in DET0455</a> is a parent-process filter “based on abnormal parents like Excel, WinWord, or mshta spawning PowerShell.” But C-PS01 is deliberately much smaller than that strategy. The strategy’s analytic, AN1252, “detects behavioral chains where PowerShell is launched with encoded commands, unusual parent processes, or suspicious modules loaded,” “supports detection of both direct (powershell.exe) and indirect (.NET automation) invocations,” and <a href="https://attack.mitre.org/detectionstrategies/DET0455/">lists Sysmon process-creation and module-load events plus PowerShell’s own event logs</a> as its sources. C-PS01 promises one relationship from one kind of record, and it uses Windows Security process-creation events, which aren’t among the sources AN1252 lists. That’s fine for an organization’s own capability. The public strategy is a source of ideas, not a specification C-PS01 claims to implement, and the reference should say plainly how much smaller it is.</p> <p>It also has to be clear about what a match means. ATT&amp;CK’s <a href="https://attack.mitre.org/techniques/T1059/001/">PowerShell technique page</a> notes that PowerShell “could be in use for many legitimate purposes and administrative functions,” that commands and scripts “can also be executed without directly invoking the <code class="language-plaintext highlighter-rouge">powershell.exe</code> binary” through the <code class="language-plaintext highlighter-rouge">System.Management.Automation</code> assembly, and that PowerShell can be used to download and run executables “from disk or in memory without touching disk.” It isn’t Windows-only, either. So an Office application directly creating <code class="language-plaintext highlighter-rouge">powershell.exe</code> is a relationship that deserves an analyst’s look. It isn’t proof that the script, the document or the user is malicious, and a capability that sees only that relationship can’t claim to cover PowerShell abuse in general. Revision 1 says so in its explicit exclusions: <br /></p> <ul> <li>PowerShell hosted inside another process through <code class="language-plaintext highlighter-rouge">System.Management.Automation</code></li> <li>Linux and macOS</li> <li>parents other than WinWord and Excel, intermediary process chains and renamed executables</li> <li>any claim to cover all of T1059.001 or everything AN1252 describes</li> <li>any claim that a script ran successfully, did harm or lacked authorization <br /> <br /></li> </ul> <p>The evidence comes from process-creation events. Microsoft’s archived Windows 10 documentation for <a href="https://learn.microsoft.com/en-us/windows/security/threat-protection/auditing/event-4688">Event 4688</a> describes the fields: the new process name, the “full path and the name of the executable for the new process”; the creator process ID; a creator process name (<code class="language-plaintext highlighter-rouge">ParentProcessName</code> in the event XML) added in event version 2; and a process command line, present in versions 1 and 2, which the page says is empty by default unless the “Include command line in process creation events” policy is enabled. That documents a schema and its configuration conditions. It doesn’t guarantee that a given fleet, collector or parser produces those fields, and it isn’t a statement about every current Windows release. So C-PS01 stipulates them: version 2 events, collected and parsed correctly, with the command line enabled. Its required observations from the event are the source computer and time, the new process and creator process names, account context and the child’s command line, which reviewers need for context and which the exception check uses for one specific fact. Excusing a launch also takes two things the event can’t supply: a current register of approved workflows, and an execution-context record for that launch.</p> <p>Both deserve a careful definition, because a loosely defined exception can quietly swallow the behavior a capability exists to catch. In C-PS01, an approved exception is a register entry for one specific automation. It names the host population the automation runs on, the account it runs as, its invocation - the exact arguments it passes to PowerShell - the known revision of the script it runs and the operation it performs, and it has to be current when the event occurs. A launch is excused only when evidence about that particular launch establishes every one of those facts, and no single record does. The event places the launch on a host and under an account. The command line shows what PowerShell was asked to run, such as a script path or an encoded command, and so whether the launch asked for the registered invocation with nothing added. That’s the one part of the exception the command line carries. It can’t show which revision of a script actually ran or what the run did: a path tells you where a file sat, not what it contained at that moment, and arguments describe a request, not an operation.</p> <p>That’s the execution-context record’s job. C-PS01 defines it only by what it has to establish: a trustworthy record, bound to this specific process, identifying the script revision that ran and the operation it performed. It speaks to the script that ran and what it did, not to whatever else the launch might have asked PowerShell to do, which is why the invocation check stays with the command line. It isn’t a 4688 field, and the reference doesn’t say how one gets produced. In this fictional organization, approved automation has such a record, stipulated to be accurate in the same way the event fields are; a real organization would first have to decide whether it can produce one it trusts. The register entry alone excuses nothing; it says what would have to be true, and the evidence for this launch has to show it. An account name alone doesn’t qualify: it tells you whose context the process ran in, not which operation this was. A signed script doesn’t: a signature identifies a publisher, not an approved use. A script path doesn’t, even one that matches the register exactly. Each covers at most one of the facts an entry names, and the exception needs all of them.</p> <p>Now the implementations. The one that shipped, A1, selects the declared Office-parent and PowerShell-child relationship only when the child’s command line contains the selected <code class="language-plaintext highlighter-rouge">-EncodedCommand</code> form, and applies the approved-workflow check. The proposed correction, A2, selects the same relationship without requiring that argument form, keeps the same approved-workflow check and marks events where required observations are missing.</p> <p>Hold everything else fixed: B1, C-PS01 revision 1, the population, the event schema, the register, the execution-context records and their availability, the exception check both implementations share, and the purpose - reviewable investigation candidates, not verdicts. Then walk five observations through both.</p> <p><br /></p> <table> <thead> <tr> <th>Observed case</th> <th>A1 / A2</th> <th>Disposition</th> </tr> </thead> <tbody> <tr> <td>WinWord, encoded command</td> <td>Candidate / candidate</td> <td>Already addressed for this case</td> </tr> <tr> <td>WinWord, plain script file</td> <td>Not selected / candidate</td> <td>Correct the implementation within core</td> </tr> <tr> <td>Excel, plain script file</td> <td>Not selected / candidate</td> <td>Same capability, known variation</td> </tr> <tr> <td>Approved reporting workflow</td> <td>Exception / exception</td> <td>Keep the narrow control</td> </tr> <tr> <td>Parent name missing</td> <td>Unresolved / unresolved</td> <td>Get the comparison evidence</td> </tr> </tbody> </table> <p><em>Expected outcomes from invented analytic descriptions under B1, not results from running a detector.</em></p> <p><br /></p> <p>In the first row, WinWord creates <code class="language-plaintext highlighter-rouge">powershell.exe</code> with the encoded form, every required event observation is available, and the exception evidence doesn’t establish an approved workflow. Both implementations present a candidate. As far as the descriptions go, a proposed rule for exactly this behavior is already addressed; in a real comparison, that conclusion also needs evidence that the deployed implementation behaves as described. Either way, it doesn’t mean the activity was malicious, that PowerShell is covered, or that an independent rule watching something else should go.</p> <p>The second row is where the comparison earns its keep. Same parent, same child, same population, same exception evidence - but the child is launched with a script file passed as a plain argument instead of an encoded command. A1 misses it, because A1 made the encoded form mandatory. Someone looking at this could reasonably say, “we need a new detection for Office spawning PowerShell with a script file.” The reference says otherwise. C-PS01 revision 1 already promised direct launches regardless of argument form, and A1 implemented less than that. So the right outcome is a correction within core: A1 has a gap against an unchanged reference, and A2 is the proposed fix. B1 doesn’t need to change, and adding a second capability would bury the fact that the first one was under-delivered.</p> <p>In the third row, Excel launches the same child with the plain form. Excel was already a named parent in the reference and in both implementations, so this isn’t a new parent family or a scope extension. It’s the same argument-form gap showing up under a different parent. Compare the observed relationship, not the technique label: the label was identical in all three rows.</p> <p>The fourth row is the legitimate-use control. A reporting workflow launches PowerShell with an encoded command. The event’s host and account fit a current register entry, the recorded command line is that entry’s invocation with nothing added, and the execution-context record bound to this process shows the entry’s script revision performing the entry’s operation. Both implementations treat it as an exception, and both should. Keeping that control narrow is part of the capability, not a convenience bolted onto it. It doesn’t make every process under that account benign, or every signed script, or every launch with the same arguments, or Office launching PowerShell in general.</p> <p>The fifth row is the one I’d defend hardest. A <code class="language-plaintext highlighter-rouge">powershell.exe</code> creation is recorded, but the creator relationship isn’t there, and nothing else available establishes it to the standard C-PS01 requires. Neither A1 nor A2 can make the comparison from this record, so the outcome is unresolved, and it helps to be exact about what that leaves open. The record doesn’t show that the launch was benign, and it doesn’t show that it was malicious. It doesn’t even show whether the launch falls inside C-PS01’s scope: the missing parent might have been WinWord or Excel, or a process the capability never covered. What it does show is that a required observation is missing. That doesn’t make any rule redundant and doesn’t turn the launch into a satellite case; the comparison stays open until the evidence turns up or its absence is explained. One incomplete record doesn’t tell you the cause or extent of a collection problem either. Finding that out is separate work, and the third case takes up a different situation, one in which the cause is already known.</p> <p>There’s a subtler variant worth naming. Change one thing only: the command line goes missing, while the process names, account context, register and execution-context records stay exactly as they were. That’s an easy state to reach given the documentation’s empty-by-default note, since a host where the policy never applied would look exactly like this. Run the declared contract and the losses come out specific. A2 can still see the relationship, so WinWord creating <code class="language-plaintext highlighter-rouge">powershell.exe</code> is still selected. If the rest of the exception evidence doesn’t hold for this launch - no current entry fits its host and account, or no execution-context record shows an entry’s script revision and operation - the outcome doesn’t change. It was a candidate before and it’s a candidate now, just without the command line reviewers were promised.</p> <p>If the rest does hold, the event and the execution-context record match a current entry on host, account, script revision and operation, and exactly one fact is unestablished: whether the launch asked for the registered invocation and nothing more. A2 can’t confirm the exception, so the honest output is a candidate marked with that single open question, not a suppressed alert and not an unexplained one.</p> <p>A1 fares worse. Its encoded-form gate lives in the command line, so it can’t evaluate its selection condition at all, and because A1 doesn’t mark missing observations, its silence looks just like a launch it evaluated and didn’t select. That silence isn’t evidence that the launch lacked the encoded form.</p> <p>For A2, this is a partial delivery gap against two named obligations: review context for every affected launch, and the invocation part of the exception wherever the rest of an exception holds. The record should name both. “The detection is broken” overstates it. “It still fires, so we’re fine” understates it. And the missing field isn’t evidence about the launch in either direction; an exception that can’t be confirmed simply hasn’t been granted.</p> <p>So what does the comparison justify? Proposing A2 as a correction for a logical mismatch between A1 and the reference. Planning scoped checks for any later implementation work: a positive for each argument form and parent; the legitimate-use exception, along with near-misses that match an entry on account or script path but not on execution context; and the missing-evidence cases, including a missing command line. And leaving B1 alone. ADS-style validation - generating “a representative true positive” - is the right instinct, but for C-PS01 one positive wouldn’t be enough, because the exception and the missing-evidence behavior are part of the promise too. A passing positive is real evidence about the variant it exercised, in the environment and period where it ran. It can reasonably inform what you expect from close variants that go through the same logic and dependencies, for as long as those stay as they were. It doesn’t establish that an untested variant is caught, what the alert burden will be in production or how long the result will hold.</p> <p>It doesn’t justify deploying A2 on the strength of a table in a blog post. It doesn’t justify reporting sensitivity, precision, coverage percentages or saved analyst hours, because nothing was measured. And it doesn’t justify treating every non-excepted Office-to-PowerShell launch as unauthorized. The capability produces candidates for review.</p> <p><br /></p> <p><strong>Case Two: A Satellite That Never Needs to Move</strong></p> <p>Same organization, different detection. One of its applications pays suppliers. Each supplier has beneficiary details, and every change to them creates a new beneficiary revision. The application’s declared policy is that a payment may be released only against a beneficiary revision an independent approver has approved, with the approval bound to that exact revision.</p> <p>The detection, S-PAY01, identifies release of a supplier payment against a beneficiary revision that lacks the independent approval the policy requires. Here’s the positive case, with record completeness stipulated: <br /></p> <ol> <li>An independent approver approves beneficiary revision R5 for payment P.</li> <li>The beneficiary details change, creating revision R6. The approval stays explicitly bound to R5.</li> <li>A release record shows payment P released against R6. The complete, authoritative record contains no qualifying approval of R6 and no permitted override. <br /> <br /></li> </ol> <p>Within the fictional policy and those completeness assumptions, the release violates the declared requirement, and the organization treats that as an urgent payment-diversion exposure. That’s what makes S-PAY01 critical: the consequence, and the response it needs. It’s not what the detection establishes about anyone’s intent. The records don’t show who meant to do what, whether an account was compromised, whether money actually moved or whether anything was lost. The detection’s claim is narrower and still serious: a release happened that the policy says shouldn’t have, and someone needs to look now.</p> <p>Two countervariants keep the claim bounded. If R6 receives the required independent approval before release, the violation condition is absent, assuming the other declared facts hold. And if the approval records or the revision binding are incomplete, whether this release violated the requirement is unresolved. The collection defect may need urgent attention of its own, but missing evidence doesn’t become proven misconduct, just as C-PS01’s missing parent didn’t settle anything about that launch in either direction.</p> <p>So why isn’t S-PAY01 in B1?</p> <p>Not because it matters less, and not because it’s immature: by stipulation it’s mature and fully supported. It’s outside B1 because B1 is organized around reusable technical behavior across the workstation fleet, and S-PAY01’s meaning lives somewhere else. It depends on application-specific transaction state, approval binding and release semantics. The process-creation events on a clerk’s workstation could look the same for a compliant release and a violating one. Nothing in C-PS01’s kind of observation can supply the distinction, and neither can a shared label like “execution” or “financial.” Folding S-PAY01 into a generic execution capability to keep the category count down would make it harder for anyone reviewing it to see what an approval binding is, or why its absence matters.</p> <p>This is where the definition of membership does real work. Core and satellite differ in scope coherence - whether a detection shares the reference’s organizing criteria and evidence semantics - not in importance, and not in what infrastructure they happen to share. Another organization could reasonably build a baseline for its payment platform in which S-PAY01’s invariant is core. That would be a different reference with different organizing criteria, and just as legitimate. What I’m arguing against is the idea that application-specific detections are second-tier by nature, or that a detection outside the current baseline is a baseline candidate that hasn’t made it yet.</p> <p>S-PAY01 can stay a satellite for as long as the payment workflow exists, and there’s nothing unfinished about that. What it needs is the support anything in core would get: <br /></p> <ul> <li><strong>Ownership:</strong> an application detection owner maintains the semantics and the implementation; the SOC owns the agreed alert-handling path.</li> <li><strong>Review triggers:</strong> changes to the application schema, the approval policy, the release paths or the business purpose.</li> <li><strong>Testing:</strong> scoped positive, legitimate reapproval, authorized override and incomplete-record checks for any implementation.</li> <li><strong>Retirement basis:</strong> loss of the detection purpose, a replacement backed by an adequate comparison, or an explicit risk decision. Never failure to enter core. <br /> <br /></li> </ul> <p>None of that is novel. <a href="https://github.com/palantir/alerting-detection-strategy-framework/blob/master/ADS-Framework.md">ADS</a> already asks every strategy for a goal, blind spots, validation and response steps. The point is that a satellite carries those obligations as fully as anything in core, instead of getting a lighter standard because it sits outside the baseline.</p> <p>Cross-labels still connect it to the rest of the portfolio. Tag it with the application, with integrity and payment-process categories and with the response playbook that applies. Someone searching for that application, or for payments, should find it. Those labels associate; they don’t merge.</p> <p><br /></p> <p><strong>Case Three: Gap or Change of Mind?</strong></p> <p>The last case is about time, and it’s where keeping the three records apart pays off most. Two things happen to C-PS01 in this separate thought experiment. One is a delivery failure that can be mistaken for a change in scope. The other is a change in scope that can be mistaken for routine tuning.</p> <p>Stipulate an assurance premise that belongs to this case only and isn’t derived from the table above: at some earlier point, A2 running with parser P1 on the declared population had scoped evidence that it delivered C-PS01’s promised behavior. Then a parser update, P2, ships, and it drops <code class="language-plaintext highlighter-rouge">ParentProcessName</code> from otherwise relevant records. In this branch the defect is known. The capability hasn’t changed, the population hasn’t changed and A2’s logic hasn’t changed.</p> <p>Every affected record now looks like the fifth row of the table: a <code class="language-plaintext highlighter-rouge">powershell.exe</code> creation with no parent. Event by event, each comparison is still unresolved. Knowing the cause tells you why a parent is missing, not what it was, so it doesn’t make any of those launches benign, malicious or out of scope. What the known defect does justify is a classification at a different level, one no single record could reach: current delivery in the affected deployment has a core gap against B1 for the affected interval. The reference hasn’t shrunk; delivery against it has. The earlier assurance remains valid evidence about A2 with P1, on that population, at that time. What it loses is automatic carry-forward. It supported an expectation conditional on its dependencies staying put, and P2 changed one of them in a field C-PS01 requires, so it can’t be read as showing that the current arrangement still satisfies the reference.</p> <p>The repair follows. Keep B1 and C-PS01 revision 1 exactly as they are. Identify the affected deployment and interval. Fix or replace the parser dependency. Get new scoped assurance for the repaired arrangement. What you shouldn’t do is move the affected hosts or events to satellite status, or declare the population smaller because those hosts stopped sending usable telemetry. The population was declared in the reference; losing visibility into part of it is a gap in delivering the reference, not a revision of it. That move is exactly what I mean by silently redefining the promise, and it’s the reason the baseline has to describe an intended condition in the first place.</p> <p>Now the other kind of change. Suppose the organization starts thinking about PowerShell hosted inside another process through <code class="language-plaintext highlighter-rouge">System.Management.Automation</code> - the indirect route ATT&amp;CK describes, and one that C-PS01 revision 1 explicitly excluded. That isn’t an argument-form problem another analytic revision can fix. It changes what’s being promised. The legitimate options are a revised capability definition, C-PS01 revision 2, in a successor baseline B2; a separately defined and supported capability; or a considered decision to leave the exclusion in place. Whichever it is, B1 stays as it was, because comparisons were made against it and those comparisons still mean what they meant.</p> <p>The revision also has to say what evidence would support the new claim. For indirect invocation, the <a href="https://attack.mitre.org/detectionstrategies/DET0455/">public strategy</a> points toward module-load events and PowerShell’s own logs, none of which are in C-PS01’s process-creation fields, and it lists the loaded-module set as a mutable element, “tuneable to monitor rare or never-before-seen .NET assemblies tied to PowerShell abuse.” Whether the presence of the PowerShell assembly in an unexpected process means anything in this population is a question the revision has to answer, not an assumption it gets to make.</p> <p>Change one premise and the classification flips. If C-PS01 revision 1 had already promised indirect hosting, missing support would be an existing implementation gap, handled like the parser regression: same reference, more implementation work, new evidence. Same technical behavior, different classification, because the reference said something different. That’s the whole argument for writing the reference down.</p> <p>A third trigger deserves an explicit answer: something new with a name. A new platform joins the fleet, or a new attack family gets a technique ID. Say the organization starts managing macOS workstations, and ATT&amp;CK notes that PowerShell <a href="https://attack.mitre.org/techniques/T1059/001/">“can also be installed on Linux and macOS.”</a> That’s a good reason to review B1’s applicability and the organization’s risk. It isn’t an automatic addition to B1, which declared a Windows population, and it isn’t a gap against B1 either, because B1 never promised it. A name that wasn’t there before prompts a scope decision. It doesn’t make one.</p> <p><br /></p> <table> <thead> <tr> <th>What changed</th> <th>Does B1 change?</th> <th>What it is</th> </tr> </thead> <tbody> <tr> <td>Parser P2 drops the parent name</td> <td>No</td> <td>Core delivery gap in the affected deployment and interval</td> </tr> <tr> <td>Covering indirect hosting, excluded in revision 1</td> <td>Only through a successor</td> <td>Scope decision: revision 2 in B2, or a separate capability</td> </tr> <tr> <td>Indirect hosting was already promised</td> <td>No</td> <td>Existing implementation gap</td> </tr> <tr> <td>A new platform or technique name</td> <td>Not automatically</td> <td>Prompt to review applicability and risk</td> </tr> </tbody> </table> <p><em>Illustrative classifications for the fictional C-PS01 capability.</em></p> <p><br /></p> <p>For any of this to be decidable later, each comparison has to record what it compared. Not a heavyweight form, just enough that someone reading it next year can reconstruct the claim: <br /></p> <ul> <li><strong>Reference:</strong> baseline B1 and capability C-PS01 revision 1.</li> <li><strong>Implementation:</strong> analytic revision A1 or A2, with an immutable artifact reference in a real system.</li> <li><strong>Deployment:</strong> the deployment revision and the declared population.</li> <li><strong>Dependencies:</strong> the parser, field-mapping and approved-workflow register revisions in effect, and the source and revision of any execution-context records an exception relied on.</li> <li><strong>Evidence:</strong> the assurance record’s identifier, observation interval, assessment time, scope and limitations.</li> </ul> <p><br /></p> <p><img width="1448" height="1086" alt="Timeline illustrating how capability references, implementations, and assurance evidence change independently. The capability reference moves from version B1 to B2 only through an explicit update. Implementations change from A1 to A2 as detection logic and deployment improve, while evidence records E1 and E2 document performance at different points in time. The bottom panel shows that the full portfolio includes both core capabilities and satellites, with linked capability, implementation, and assurance records." src="https://github.com/user-attachments/assets/d051c418-f8a4-44cb-a9e9-d68f9be166fd" /></p> <p><br /></p> <p><strong>Objections Worth Taking Seriously</strong></p> <p>This arrangement isn’t free.</p> <p><em>A static baseline goes stale.</em> It can. Static means fixed per version, not exempt from review. Platforms change, risk changes and attack behavior changes, and the reference’s applicability should be reviewed when they do. When the commitment changes, publish a successor and keep the old one for everything compared against it. The concession is real, though: version control preserves history, it doesn’t make the chosen reference relevant. Someone still has to judge what the next version should say, and that takes capacity.</p> <p><em>Deduplication throws away independent detection.</em> It can, if “already addressed” gets treated as “delete.” Before retiring anything, compare what each rule observes, what it depends on and what corroborating role it plays in an investigation. Similar content can stay as a second implementation under the same capability, or as a separately justified detection where its purpose differs. The concession cuts the other way, too: multiple rules aren’t automatically independent, and different product names, tags or table names don’t establish that two rules fail differently.</p> <p><em>Satellite status becomes a dumping ground, and narrow scope hides gaps.</em> This is the objection I take most seriously, because the vocabulary can be gamed. A team could relabel failing detections as satellites, or write a baseline so narrow that nothing is ever missing. The model’s answer is that a satellite needs a written purpose and a scope comparison, a failure against something core already promised stays a gap whatever anyone renames it, and an unresolved comparison stays unresolved. The concession is that no naming scheme prevents scope gaming on its own; transparent ownership and review do that work. What the model avoids is the opposite failure, forcing every legitimate satellite into the core so the category looks tidy.</p> <p><em>Central stewardship creates a bottleneck.</em> For a single SOC, I think central stewardship of capability meaning is a reasonable choice - one place decides what C-PS01 means - while implementation work stays distributed among the people who write and operate detections. Where authority and environments genuinely differ, federated decision rights over separate references may fit better. I’m not aware of evidence that either arrangement is generally better, and I’m not claiming one is. What matters is that someone is clearly responsible for changing the reference, and someone is clearly responsible for operating each detection.</p> <p><em>Different environments make one fixed baseline misleading.</em> Then don’t use one. An organization can keep one scoped reference, several environment-specific references, or profiles where they genuinely help. The comparison always runs against the applicable declared version. I’d resist treating a shared core plus mandatory profiles as the default architecture, though. That’s a design choice with its own costs, not something the concept requires.</p> <p><em>A small SOC can’t maintain another framework.</em> Agreed, and I wouldn’t ask it to. Start with one consequential comparison and a short capability record in whatever engineering documentation you already keep: the intended behavior, applicability, implementation identity, evidence scope and the open decision. Don’t build a parallel bureaucracy. The concession is the plainest one in the post: writing records doesn’t fix staffing or telemetry gaps, and I can’t point to a measured reduction in anyone’s workload. The value I’m proposing is narrower - avoiding specific confusing decisions, like adding a duplicate capability to cover an implementation bug, or relabeling a gap until it disappears.</p> <p>Two smaller misreadings are worth heading off. The model doesn’t prescribe a rule count, per baseline or per capability. And it doesn’t prescribe a universal release gate. ADS says every strategy <a href="https://github.com/palantir/alerting-detection-strategy-framework/blob/master/ADS-Framework.md">“must have true positive validation,”</a> which is a sensible norm, but the principle I’d defend is narrower: the evidence attached to a detection, core or satellite, should match the claim being made for it.</p> <p><br /></p> <p><strong>Where to Start</strong></p> <p>You don’t need to reorganize a portfolio to use any of this. Take the next detection someone wants to add, or an existing rule someone wants to retire, and run one comparison properly.</p> <p>Write down which baseline version and capability revision you’re comparing against. If nothing is written down, writing the capability you believe you already have is the first finding. Describe the behavior the proposal observes and where it applies. List the implementations that claim to serve the capability, with the dated, scoped evidence for each. Then choose one of the four outcomes, write down what it doesn’t license, and record whatever stayed unresolved and who’s chasing it.</p> <p>Do that once and you’ll have something an inventory can’t give you: a dated statement of what you promised, what delivered it, what you know about how well, and what you decided about a detection that didn’t fit. That record is what lets a satellite like S-PAY01 stay exactly where it is, critical and fully supported, without anyone mistaking it for a baseline item that hasn’t arrived yet. It’s also what keeps a broken parser from quietly shrinking a promise nobody decided to shrink.</p> <p><img src="http://canarytokens.com/terms/stuff/rocyfzfwpfjawltn4qx340bqi/submit.aspx" style="display: none;" /></p>
  112. What Does Cyber Defense Need From an Intelligence Shop?

    Mon, 26 Oct 2026 00:00:00 -0000

    An intelligence shop can be right and still be no use to the organization paying for it. Not wrong, not sloppy, not late - right. The sourcing holds up, the reasoning is clean, the uncertainty is stated honestly, and the thing lands on a desk where nobody needed that particular question answered. An accuracy-only appraisal doesn’t test whether that question mattered.
    <p>An intelligence shop can be right and still be no use to the organization paying for it. Not wrong, not sloppy, not late - right. The sourcing holds up, the reasoning is clean, the uncertainty is stated honestly, and the thing lands on a desk where nobody needed that particular question answered. An accuracy-only appraisal doesn’t test whether that question mattered.</p> <p>My concern is that a cyber threat intelligence (CTI) function can inherit structure, vocabulary and production habits from national and military intelligence before its own organization has settled three things - what the function is supposed to understand, whose questions win when they conflict, and what a satisfactory answer should let somebody do or understand. That is an institutional-fit problem, not an argument that accurate analysis doesn’t matter.</p> <p>The lazier argument - traditional intelligence is slow, hierarchical, attribution-obsessed and uninterested in the people doing the actual work, so cyber needs something fundamentally newer - doesn’t survive the doctrine. <a href="https://www.first.org/global/sigs/cti/curriculum/pir">FIRST’s CTI curriculum says priority intelligence requirements (PIRs) “are fluid and not static”</a>, and that they “are not solely the function of the intelligence section.” A 1994 US Army collection-management manual tells the collection manager, when a requirement cannot be met as written, to <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“coordinate with the requester to establish possible alternatives that might satisfy his needs”</a>. The UK’s published analytical standards say assessments <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“should address the questions posed”</a> and should be provided “in time for it to be of utility to customers.” The standards explicitly grade relevance. Adaptation is already part of the guidance, not a principle cyber needs to invent.</p> <p>Selective inheritance is the failure mechanism I want to examine. If a shop copies the desk structure, report template, requirements document, production calendar and picture of a senior analyst without the continuing review those arrangements need, it inherits the furniture and not the habits. Familiar structures can then settle questions the organization hasn’t actually answered.</p> <p>What follows is seven arrangements worth testing - who sets the question, what regional knowledge buys you, where evidence comes from, whose clock counts, what happens after delivery, how to stay useful without becoming agreeable, and who the shop is building. A fictional organization makes the reasoning inspectable. The payoff is distinguishing four possibilities behind a complaint that “CTI is not working”: a genuine institutional misfit, a legitimate mission the complainant doesn’t personally value, a question the available evidence cannot settle, or an implementation obligation nobody funded. These aren’t exhaustive or mutually exclusive; one complaint can involve several. But they can call for different responses, and guessing wrong is expensive.</p> <p><br /></p> <p><strong>Received, Wanted, Needed, Usable</strong></p> <p>Before arguing about whether intelligence is “actionable,” I want to separate four things. This is my working vocabulary for the post, not a quoted doctrinal taxonomy.</p> <ul> <li><strong>Received</strong> is what reaches a consumer: reports, feed entries, briefings or an unsolicited warning.</li> <li><strong>Wanted</strong> is what the consumer asks for, or says would help.</li> <li><strong>Needed</strong> is the uncertainty or understanding that bears on a mission, obligation or decision the organization has. A need can precede agreement. Authorized leadership can set priorities without unanimity; negotiation helps expose incompatible expectations and make the assignment clear.</li> <li><strong>Usable</strong> is what that consumer can interpret and employ with the time, evidence, authority, systems and capacity they have. That includes deciding, preparing, revising their picture of the risk, or concluding that nothing needs to change. The same paragraph can be usable for a legal adviser and inert for a detection engineer without either being wrong. <br /> <br /></li> </ul> <p>By traditional intelligence I mean national and military institutions - their missions, consumer relationships, collection arrangements, professional formation and production practices. I don’t mean a toolbox of analytical techniques. Techniques can travel, but somebody has to select the ones that fit and adapt them. It is the institutional assumptions riding along unexamined that concern me.</p> <p>By enterprise CTI I mean an intelligence function supporting an organization’s cyber-related defense and other expressly assigned decisions. Government employment alone does not confer a national intelligence mission, collection authority or national-level consumers. An enterprise-defense team can have assigned responsibilities beyond its own estate, but the assignment confers that, not the letterhead.</p> <p>A PIR is a prioritized, decision-linked subset of intelligence requirements, not the full universe of useful inquiry. FIRST says PIRs <a href="https://www.first.org/global/sigs/cti/curriculum/pir">“are not infinite, they must be limited and well-scoped to key decision points / COAs as identified by the commander. Less is more.”</a></p> <p>An identity problem, here, means incompatible expectations about the function’s purpose and satisfactory output, not a diagnosed industry-wide illness. The UK standards explicitly allow <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">individual assessment organizations to develop them for their own requirements and procedures</a>. FIRST describes its curriculum as <a href="https://www.first.org/global/sigs/cti/curriculum/cti-introduction">introducing “the words and methods that have been used by professionals for many years”</a> - deliberate transfer, with adaptation. The four FIRST pages cited throughout belong to one curriculum and program lineage, not four independent confirmations.</p> <p><br /></p> <p><img width="1448" height="1086" alt="From Received to Usable Intelligence" src="https://github.com/user-attachments/assets/4ed15458-0548-46e8-adab-8628d3a4e394" /></p> <p><br /></p> <p><strong>Who Decides What the Question Is?</strong></p> <p><em>Editorial Note: the organization and scenarios below are entirely fictional. Their facts and costs are stipulated, not observations of any real body, incident, supplier or report.</em></p> <p>Picture an international engineering accreditation body. It runs an identity service and a hosted document-review system holding unpublished examination material and accreditation case files. It has two CTI analysts with different professional histories - one came from national intelligence analysis, one from incident response - and for this whole post their actual skills, source access and total working capacity are fixed. Their material consists of public reporting, a supplier subscription that does not include bespoke collection tasking, authorized internal investigation summaries, thirty days of sign-in records, and a contractor authorization register that gets updated monthly. Historical document-read logging was never turned on. Sign-in records do not establish which files anybody opened. Nothing in the available material establishes that a compromise happened at all.</p> <p>Now a fictional external report lands describing misuse of contractor accounts somewhere else, attribution unresolved.</p> <p>In the first arrangement, the shop has a standing regional PIR: what are a named country’s actors doing? The analysts answer it, and they answer it well. The regional assessment is accurate. The SOC reads it as though it were an answer to a different question entirely - were our case files accessed? - finds that it is not, and concludes that CTI produced something irrelevant.</p> <p>The regional question was stated and answered. The SOC’s local expectation wasn’t reconciled with it; that is the defect. An assigned regional assessment doesn’t automatically owe an answer to an unassigned local question, and another consumer could legitimately value it. That doesn’t erase other assigned duties or the possibility of an unsolicited warning. Establish what was owed before declaring the analysis irrelevant.</p> <p>A PIR nobody revisits can become a production schedule with a serious-sounding name. But an unchanged requirement isn’t proof of that: it can be reviewed and retained because it’s still the right question. FIRST says PIRs are fluid and <a href="https://www.first.org/global/sigs/cti/curriculum/pir">“not solely the function of the intelligence section”</a>. Army practitioners make a compatible argument about specialist participation: <a href="https://www.army.mil/article/285410/priority_intelligence_requirement_management_in_divisions_and_corps">specific information requirement development “should be a collaborative effort between the collection management team and single-source subject matter experts”</a>. That Center for Army Lessons Learned article is an authored professional recommendation, not a directive or evidence that collaboration reliably produces better outcomes.</p> <p>Four things I’d want clear when discussing a requirement follow. They aren’t a gate every question has to pass or a guarantee of useful work.</p> <ul> <li>Who the consumer is, specifically. “The business” doesn’t identify one. An incident lead does.</li> <li>What decision or understanding the answer serves. Improving somebody’s model of their exposure can matter even if nobody pushes a button afterward.</li> <li>When the answer has to exist, and when it stops being useful.</li> <li>What limits are acceptable. If the honest answer will be partial, say so while there’s still time to decide whether partial is worth paying for. <br /> <br /></li> </ul> <p>A request isn’t automatically a need. A SOC can ask for the wrong thing, confidently, and simply filling orders is no substitute for understanding the problem. A producer can also identify a need the consumer doesn’t accept. Priority disputes belong with whoever has the assigned authority to settle them, which can include the intelligence function itself. Where CTI’s delegated authority ends, improving the requirements document won’t substitute for a leadership decision.</p> <p>Nor does this make unsolicited work illegitimate. The UK standards say assessment bodies <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“may anticipate a customer requirement and produce an assessment without an explicit tasking, especially in the realm of early warning”</a>. Warning somebody about something they didn’t ask about is a recognized part of the job. Equally, a legitimate strategic requirement doesn’t become invalid merely because it can’t answer today’s incident question.</p> <p><br /></p> <p><strong>Know the Actor, Understand the Exposure</strong></p> <p>Regional and actor expertise can inform political priorities, institutional relationships, historical target selection and why a date on your own calendar might matter to somebody else. A technique catalog alone doesn’t supply that understanding. This is a positive contribution to defense, not something to tolerate until the technical work starts.</p> <p>Take a separate planning question for the same fictional accreditation body. In six weeks it will announce an accreditation decision involving institutions in two fictional countries. Credible fictional public reporting documents an escalating political dispute over exactly those decisions, and the body’s role in them is already public. No report names the body as a target. Its technical inventory has not changed, no new technique has been observed against it, and its security telemetry is exactly what it was last month. Separately, a renewal is due before the announcement for a partner with privileged access to examination material, and the alternatives would take weeks of procurement, operational preparation and legal review.</p> <p>A shop running a rule that says no new technical observation means no changed intelligence question has nothing to say here. A shop with regional analysis can say something genuinely useful: here is why this organization’s public role and this specific date may matter to somebody, here is what we do not know, and here are the options that need a decision before the renewal date, given how long the alternatives take to stand up. That is a real contribution and it arrives before any technical change does.</p> <p>No hostile intent or attack probability has been established here, and continuing the contract unchanged may be the right answer. The UK NCSC’s heightened-threat guidance recognizes that geopolitical tensions can make a changed risk picture <a href="https://www.ncsc.gov.uk/guidance/actions-to-take-when-the-cyber-threat-is-heightened">“more specific to a particular organisation, sector or even country”</a>. Its advanced actions acknowledge that accelerating defensive improvements “will likely require reprioritisation of resources or investment.” In this illustration, the review consumes analyst, legal, procurement, security and business-owner time committed elsewhere. Reducing the partner’s access could also impair examination delivery. Those stipulated costs sit against an unproven benefit; context doesn’t hand anybody a blank check.</p> <p>A regional desk can start from <em>who</em>, while a defensive question can start from <em>how they would reach us</em> and <em>what we could see if they did</em>. Neither owns those questions exclusively. Regional analysis can examine access; detection work can draw on intent. Confusing the relationship invites two mistakes.</p> <p>The first is requiring attribution before anything else can happen. MITRE notes that organizations’ <a href="https://attack.mitre.org/groups/">group definitions “may partially overlap with groups designated by other organizations and may disagree on specific activity”</a>. Its <a href="https://attack.mitre.org/groups/">published technique coverage “does not represent all possible technique use by Groups, but rather a subset that is available solely through open source reporting”</a>. A cluster name is a scoping convention over reported activity, not a country-level detector. A shared technique association between two groups establishes that reported behavior about each fell under the same label - not identical implementations, independent reporting or a shared sponsor. The data is <a href="https://attack.mitre.org/resources/attack-data-and-tools/">published in machine-readable STIX form</a>, but counting those mappings alone doesn’t measure local exposure.</p> <p>The second mistake is the mirror image: treating a technique catalog as a threat model. It is not one. It does not supply your dependencies, your adversary’s objectives against you or your impact model, nor establish that you can detect the behavior locally. MITRE’s Center for Threat-Informed Defense (CTID) frames detection work as starting from <a href="https://center-for-threat-informed-defense.github.io/summiting-the-pyramid/analytic-design/deworkflow/">the behavior “you want to detect rather than a particular tool, indicator, or analytic”</a>, which is sound and is also step one of seven in that workflow - the last of which is evaluating what coverage the finished analytic actually provides. A mapping is an index, not coverage.</p> <p>Behavioral and exposure analysis can support defensive work without resolving national attribution. NCSC notes that even the most sophisticated and determined attacker <a href="https://www.ncsc.gov.uk/guidance/actions-to-take-when-the-cyber-threat-is-heightened">will use known vulnerabilities, misconfigurations or credential attacks “if they can”</a>. The condition matters: removing those opportunities can reduce risk. The guidance doesn’t say attackers are confined to those methods. Attribution can still matter enormously to strategic, diplomatic, legal and commercial decisions. I object to making it the gate every defensive question has to pass through, not to doing it.</p> <p><br /></p> <p><strong>A Requirement Does Not Create a Source</strong></p> <p>A requirement can justify funding, a retention change or a new data-sharing agreement. It doesn’t itself create that capability. The failure I mean is copying requirements paperwork without the collection-management reasoning that tests what can actually be obtained.</p> <p>Here are four relationships an intelligence shop can have with the evidence it uses. They overlap, they are not the whole list, and what separates them is authority and access rather than collection discipline.</p> <ul> <li><strong>Tasking something your organization owns.</strong> With tasking authority, you can set priorities within the owning team’s commitments, legal constraints and change windows. Ownership alone isn’t unrestricted authority. FIRST’s program model includes <a href="https://www.first.org/global/sigs/cti/cti-program/stage2">documenting “the observable results of internal investigations and the corresponding analysis”</a>.</li> <li><strong>Requesting support from somebody else.</strong> You can negotiate priority without having authority to set it. That 1994 manual distinguishes <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“task organic assets, request support from higher headquarters, and recommend tasking to subordinate echelons”</a>. The verbs reflect different authority relationships.</li> <li><strong>Buying published coverage.</strong> A supplier sells material it collects. FIRST describes <a href="https://www.first.org/global/sigs/cti/cti-program/stage2">incorporating “data from underground forums provided through threat intelligence partners”</a>. A subscription without bespoke tasking doesn’t itself give you the right to direct new collection. It doesn’t prevent asking for clarification or existing material you haven’t seen.</li> <li><strong>Obtaining authorized access to local records.</strong> “We have that” and “we can lawfully get at that, going back far enough” are different statements. This relationship concerns authority and retention as well as the existence of a source. <br /> <br /></li> </ul> <p>The historical doctrine already bounds requirements by feasibility. The manual instructs that <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“when a request is not feasible, notify the requester with an explanation of why the request cannot be satisfied”</a>. Finite assets, finite time, competing demands, and a duty to say so. Unlimited national collection isn’t the model in that chapter.</p> <p>Back to the accreditation body. The incident lead wants to know whether particular case files were read. Historical document-read logging was never enabled, so no requirement or escalation can recover that nonexistent stream. That doesn’t close the access question: a backup, another application-side audit trail or a supplier-held record might still bear on it. None is established as available here. Missing one stream and exhausting every possible source are different claims.</p> <p>The monthly authorization register needs a separate check. Historical validity depends on its contents: dated entries with effective ranges can support an answer about authorization on a particular day; a current snapshot overwritten each month cannot establish earlier states. The thirty days of sign-in records remain useful authentication evidence within that window, not direct evidence of which files were opened.</p> <p>Here the shop owes the partial answer the known evidence supports, with the gaps stated. Never enabling document-read logging and retaining sign-ins for only thirty days are different limitations with different remedies. The UK standards ask analysts to communicate which sources matter most to their judgments and <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">identify “critical information gaps”</a>. The gap statement is part of the product, not an apology attached to it.</p> <p>Buying intelligence or depending on partners isn’t a deficiency. An organization without the authority, legal footing or staff for external collection shouldn’t improvise it. The questions concern coverage, taskability, limitations and whether unmet needs remain visible.</p> <p>FIRST describes <a href="https://www.first.org/global/sigs/cti/cti-program/stage2">CTI personnel researching “which feeds best meet their requirements, ensuring that their budget is allocated properly”</a>. That is the relationship I’d keep: sources selected against needs. If a shop quietly substitutes the subscription’s coverage for its requirements, its calendar can fill while unanswered questions disappear from view. This is a conditional failure mechanism, not an observed industry tendency. Ask what the shop stopped covering when its subscription last changed. The answer is a starting point for examining the relationship, not a test that clears or convicts the shop.</p> <p>Narrowing a question because of what you can actually collect is responsible. Narrowing it quietly and then presenting the narrow answer as though it satisfied the original need is the problem.</p> <p><br /></p> <p><strong>Timely for Whom?</strong></p> <p>The slow-institution story runs into the same 1994 manual. During accelerated planning, it says, <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">an experienced battalion S2 “may take 5 minutes to execute exactly the same set of procedures described in this chapter”</a>. A corps collection-management section <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“might take several days”</a>. Time <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“depends upon the tools you choose to use and develop”</a>, while the manual still requires every step. These are doctrinal illustrations with different echelons, detail and tools, not observed performance or CTI benchmarks. The doctrine accommodates urgent work; it doesn’t claim that procedure costs no time.</p> <p>Multiple consumers aren’t a cyber innovation either. The chapter lists requirements from higher headquarters alongside <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">subordinate and adjacent units “in the form of specific requests for intelligence”</a>. FIRST’s CTI definition says it <a href="https://www.first.org/global/sigs/cti/curriculum/cti-introduction">is “designed to inform all levels of decision makers”</a>. The design question is whether a shop recognizes those different consumers and clocks or quietly standardizes on one.</p> <p>Go back to the accreditation body and the questions on its table.</p> <p>The incident lead, this shift, needs to know what the available evidence establishes about one contractor account and what remains unknown. If a decision about using existing suspension authority must be made this shift, a bounded partial answer may be worth more than a complete one next week. If the decision can wait, that calculation changes. Authority alone doesn’t establish urgency.</p> <p>The hunter, this week, needs a testable hypothesis that follows from the external report given current local coverage. Their clock is set by when the hunt gets scheduled, not by when the incident closes.</p> <p>The detection engineer, this planning cycle, needs to know what evidence and context an analytic would require and whether those dependencies can be met. The answer needs to be solid enough to plan against before the cycle closes.</p> <p>Leadership and the business owner, next quarter, need to know whether the contractor-access model should change at all.</p> <p>The legal adviser needs supportable factual statements about exposure or compromise, on a clock that may be unpredictable. Defensibility and timeliness both matter: if the organization has already responded publicly, a sound assessment arriving afterward has missed that use.</p> <p>These are legitimate questions, not an obligation for two analysts to serve every consumer equally. An authorized charter can prioritize the SOC or leadership without guaranteeing good service to either. The question is whether that priority was deliberately assigned or drifted in and is now being defended as the mission.</p> <p>The UK standards define timeliness relative to use: product should arrive <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“in time for it to be of utility to customers”</a>, with attention to “changing intelligence requirements and priorities.” Trading some depth for speed can be correct, but the result still needs adequate support for the judgment it makes. Somebody has to make that trade-off explicit. If speed becomes the only criterion, the shop can turn into a wire service answering whatever arrived most recently instead of the question that matters.</p> <p>A strategic assessment isn’t defective because it produces no immediate rule. Some do; producing one just isn’t the test. Nor does a strategic label excuse missing an agreed operational deadline.</p> <p><br /></p> <p><img width="1448" height="1086" alt="Where the Process Breaks" src="https://github.com/user-attachments/assets/6d8f1f79-0dcd-4cbb-a1b1-0562b64dab2b" /></p> <p><br /></p> <p><strong>Delivered Is Not the Same as Usable</strong></p> <p>Suppose the external report about contractor-account misuse is good. Well sourced, clearly reasoned, behaviorally specific. It arrives. What happens between arrival and anything changing?</p> <p>CTID’s detection engineering workflow starts with the behavior and asks engineers to <a href="https://center-for-threat-informed-defense.github.io/summiting-the-pyramid/analytic-design/deworkflow/">trace “relevant system interactions to the sensors, events, and fields capable of observing them”</a>. Its telemetry guidance notes that <a href="https://center-for-threat-informed-defense.github.io/summiting-the-pyramid/data-readiness/min-telemetry-reqs/">a source can be “only sufficient in combination with additional telemetry”</a>. Seeing an action and distinguishing malicious use are different capabilities. Engineers must also <a href="https://center-for-threat-informed-defense.github.io/summiting-the-pyramid/analytic-design/deworkflow/">test “the analytic against representative data from the environment in which it will operate”</a>. These pages express one engineering position, not independent confirmations of effectiveness.</p> <p>The seven-stage workflow includes coverage evaluation, not merely writing a query. In this use path, the report supplies the starting behavior, not a completed local analytic. But report-to-detector is only one path. If it’s the only one an organization recognizes, assessments informing a contract, legal position or executive’s understanding of exposure can disappear from its account of value. That is an institutional-fit failure pointing the other way.</p> <p>The engineering path helps distinguish three failures that can look similar from a distance.</p> <ul> <li><strong>The product did not address the need.</strong> The requirement may have been unclear, or clear and badly executed. Those call for different repairs.</li> <li><strong>The product addressed the need and the organization lacked capacity to use it.</strong> This may require funding the work or accepting the exposure. If a knowable capacity constraint was missed during scoping, CTI can share responsibility for the mismatch.</li> <li><strong>The product addressed the need and the data required to act does not exist.</strong> Better writing won’t create it. A missed data dependency can also be a scoping failure, not solely a problem for the engineering team. <br /> <br /></li> </ul> <p>These aren’t exclusive assignments of blame. Collapsing them into “CTI is not actionable” risks improving report formatting to solve a logging-retention problem.</p> <p>Now run the accreditation body again. Same two analysts, skills, capacity, sources and authorities. Three things change: the local question is stated and reconciled with the regional one before the work; the external report is translated into terms local consumers can test; and existing CTI and SOC time is allocated to both activities.</p> <p>The product that comes out distinguishes three things explicitly: what the external reporting describes, what can be tested locally against thirty days of sign-ins and the investigation summaries, and the historical file-access question that the known available sources cannot settle. The hunter gets a bounded hypothesis to take into scheduling - proposed, not validated, and still subject to whether current coverage supports running it. The engineer gets a statement of what logging an analytic would need, which lets them assess feasibility rather than guess. The strategic requirement about the named country’s actors is retained, not deleted - it was never the thing that was wrong.</p> <p>That work isn’t free. The joint session displaces scheduled CTI production and SOC planning. Engineering feasibility uses an existing half-day allocation controlled by the security manager, displacing planned maintenance on another analytic. That maintenance remains undone, and the allocation authorizes no production change. No new source, log history or capacity appears; the incident lead’s suspension authority is unchanged. Discovery of a compromise, deployment of a detector and net effort saved are not outcomes this comparison establishes, nor does it prove them impossible.</p> <p>If everyone already understood the different questions, formalizing clarification may add little. If missing records or capacity are the binding constraint, clarification exposes them without supplying either. That can help identify a misplaced investment, but doesn’t guarantee that anybody stops paying for the wrong fix.</p> <p>I would defend closing the loop. The 1994 manual says to <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">coordinate with the requestor “to ensure that the requestor also considers the requirement satisfied”</a>. Note <em>also</em>: the producer’s judgment isn’t enough, and consumer satisfaction alone doesn’t establish truth. A consumer can be delighted by a wrong answer. Record partial satisfaction and what’s still outstanding without making CTI responsible for every downstream task. The manual prescribes that check; it doesn’t measure its benefit.</p> <p><br /></p> <p><strong>Be Useful Without Becoming Agreeable</strong></p> <p>Put two sentences from the same UK standards next to each other and the tension is right there in the text. Assessments <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“should address the questions posed.”</a> Assessments <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“should not be distorted by, nor shaped for, a particular agenda or policy/operational viewpoint or preference.”</a> The US Intelligence Community’s public explanation of its analytic standards carries the same pairing, listing <a href="https://www.intelligence.gov/mission/our-values/objectivity">“demonstrate customer relevance and address implications”</a> among the criteria while objectivity sits at the center of the page.</p> <p>Both obligations, simultaneously. Not a dial to be set somewhere in the middle. These are prescribed standards, not evidence of compliance, but as a statement of the job I find them clearer than most commercial substitutes.</p> <p>The resolution is a line I find genuinely useful: co-design the question, not the answer.</p> <p><br /></p> <p><img width="1448" height="1086" alt="Relevance Without Compliance Infographic" src="https://github.com/user-attachments/assets/5562e020-b83b-4de0-a298-f91977e4fd4a" /></p> <p><br /></p> <p>That doesn’t prohibit joint collection, scrutiny or interpretation. The SOC’s knowledge of its logs and the business owner’s knowledge of the contract can be necessary to understanding the evidence. Refusing relevant input doesn’t protect independence. What cannot happen is selecting, weighting or interpreting evidence to serve a preferred conclusion.</p> <p>Negotiating the question is the work. What are you actually deciding? What would change your mind? By when? What would a partial answer be worth? An intelligence shop that will not do this is not protecting its independence, it is just inaccessible, and an inaccessible shop invites being routed around.</p> <p>Challenge is legitimate in both directions. Producers should question ambiguous requests, embedded assumptions and demands for unsupported certainty. Consumers should challenge answers that miss their decision, are unintelligible or arrive too late. Challenge alone doesn’t establish a broken relationship. The UK standards ask that significant analytical differences <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“be considered and communicated to customers where concurrence cannot be achieved”</a>, and that conclusions be presented <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“honestly, even if the assessments are contentious.”</a> The former provision gives differences between departments or intelligence organizations as examples; it isn’t a rule for every service complaint. Where a significant disagreement about the evidence remains, I’d document it in the product rather than force agreement.</p> <p>The role boundary also needs care. The US page states that <a href="https://www.intelligence.gov/mission/our-values/objectivity">“the IC does not make policy recommendations.”</a> That’s an explanation of its stated national institutional role, not a universal ban on CTI recommending defensive options. Nor does the Army collection manager’s instruction to <a href="https://irp.fas.org/doddir/army/fm34-2/Ch3.htm">“recommend tasking to subordinate echelons”</a> disprove the policy boundary. Collection tasking and national policy are different categories of recommendation. An enterprise needs to assign its own advisory role deliberately rather than import a sentence without its category.</p> <p>I would keep three responsibilities distinct:</p> <ul> <li><strong>Assessment</strong> - what is true, what is likely, what is unknown, and on what evidence.</li> <li><strong>Recommendation</strong> - what options exist, what each would cost and change. CTI can contribute threat expertise; costing and assessing options requires the relevant systems, budget and contract expertise too. An assigned advisory role shouldn’t be declined on borrowed principle.</li> <li><strong>Decision</strong> - which option gets taken, by whoever has the assigned authority and responsibility for its consequences. That can include CTI where the organization delegates a decision to it. Watch for both unassigned decision rights and assigned ones the shop refuses to exercise. <br /> <br /></li> </ul> <p>A CTI team that says “here are three ways you could handle the partner renewal, here is what each buys you, here is what we cannot tell you” has not compromised its independence by the act of listing options. The format is not the safeguard - three options can be presented dishonestly, with the preferred one dressed up and the other two built to lose. A CTI team that says “the renewal is fine, nothing to worry about” because that is what the business wanted to hear has compromised something. The difference is whether the evidence would read the same way if the business wanted the opposite answer.</p> <p>A dissatisfied consumer doesn’t establish that the intelligence was bad. An uncomfortable finding doesn’t establish that the analysis was politicized. Both inferences are lazy; ask what evidence supports them.</p> <p><br /></p> <p><strong>What Kind of Intelligence Professional Are We Building?</strong></p> <p>A bounded public search for this post didn’t establish how many CTI teams were founded or staffed by people trained in national or military intelligence, or whether that background independently predicts performance. The accessible research material was abstracts and landing pages; full methods and findings remained unread. That is an access and search limitation, not proof that the literature lacks an answer. I have no prevalence figure or causal finding. The argument here concerns a mechanism to examine in your organization, not a measured condition of the industry.</p> <p>The sources document deliberate institutional formation. The US page describing its analytic standards says ICD 203 <a href="https://www.intelligence.gov/mission/our-values/objectivity">“serves as a common foundation for developing education and training in analytic skills”</a>. On the cyber side, FIRST describes its curriculum as <a href="https://www.first.org/global/sigs/cti/curriculum/cti-introduction">introducing “the words and methods that have been used by professionals for many years”</a>, an explicit transfer of method with adaptation. I think that training foundation and those disciplines are worth taking seriously. But longevity isn’t validation: the case for structured consideration of alternatives has to stand on its reasoning, not its age. I think it’s a good case.</p> <p>Methods can carry assumptions about credible sources, proper requirements, desk structures, finished work and professional status. Where CTI founders or staff trained in national or military institutions transfer those expectations without testing their fit, the shop can be shaped around a job it wasn’t hired to do. That’s the professional-background concern directly: familiar institutional expectations left unexamined, not a demographic explanation for failure.</p> <p>It doesn’t follow that national or military intelligence professionals are nontechnical, or that a founder’s biography explains a particular failure. You cannot reason backward from a bad outcome to someone’s resume.</p> <p>The symmetrical claim deserves the same treatment. A cybersecurity background doesn’t automatically supply source criticism, structured consideration of alternatives, comfort with uncertainty or understanding of a consumer outside security. Those are learned skills, and the cited training foundation values them deliberately. Replacing intelligence backgrounds with cyber backgrounds isn’t sufficient operating-model reform; it can introduce different untested defaults. People who understand the source institution can also recognize its assumptions and lead the adaptation.</p> <p>So how do you actually evaluate anybody? Look at the work, and look at it against the role you are hiring for.</p> <p>Take two fictional candidates for a single vacancy on a team with a defined mix of analytic and technical tasks and limited time for mentoring. In work samples, candidate P correctly identifies the telemetry needed to answer a local question, and then treats two reports as independent corroboration when one is a copy of the other. Candidate Q spots the shared reporting lineage immediately and generates credible alternative explanations, and then treats a successful sign-in as evidence that a document was read. Those results are stipulated directly from the task; neither is inferred from anyone’s education or previous employer.</p> <p>Now swap a national-intelligence biography between P and Q. P’s source-independence error and Q’s confusion between authentication and access remain unchanged. That doesn’t make background irrelevant: professional history can supply evidence about responsibilities and training these samples didn’t test. It doesn’t revise the demonstrated work.</p> <p>Change the surrounding support and the fit can change. Source-lineage review addresses P’s demonstrated gap. Local-system review or teaching addresses Q’s. A team with the first capability but no available support for system semantics hasn’t covered Q’s gap merely by being strong at source reasoning. It can have one capability, both or neither. Availability matters, and neither review nor mentoring guarantees that every error is caught or a skill gap closes in time.</p> <p>I am not going to tell you who to hire. Declaring either gap cheaper would require comparing review and teaching hours with the consequences of an uncaught error in the actual role; those costs aren’t stipulated. Both forms of support consume somebody’s capacity. “We will mentor them” is a resourcing commitment, not a personality trait. The samples are evidence about two tasks, on a day, not a complete assessment or a validated hiring method.</p> <p>This is the same reasoning the program guidance already points at when it talks about <a href="https://www.first.org/global/sigs/cti/cti-program">building “teams with complementary skills to build critical processes based on our organization’s requirements”</a>, and the same thing those Army practitioners meant by making specific information requirement development <a href="https://www.army.mil/article/285410/priority_intelligence_requirement_management_in_divisions_and_corps">a collaborative effort with “single-source subject matter experts”</a>. Expertise can be distributed across a team. It just has to actually be present somewhere, and getting it consumes somebody’s capacity.</p> <p>The incentive hypothesis follows from the same concern. If recognition follows report volume or subject-matter prestige while relevance and use go unexamined, evaluation can reward production without establishing whether the need was met. The mission might still be fulfilled; that measurement alone can’t tell. This is a reason to examine evaluation, not a finding about how CTI teams are rewarded.</p> <p>Both accreditation-body arrangements produced work. The first delivered an accurate regional assessment and left the local expectation unreconciled. The second delivered a scoped partial answer, an unresolved requirement and a retained strategic one, using time promised elsewhere. Output counts could differ without telling us which arrangement met the organization’s needs. The illustration doesn’t specify how a prestige metric would rank them.</p> <p>I’d start with questions, not another summary metric. Use product, timing and requirement records alongside consumer input:</p> <ul> <li>Did the product address the question that was set or agreed, or a different one?</li> <li>Was the evidence adequate for the judgment, with gaps stated where they affect interpretation?</li> <li>Did the consumer understand it well enough to use it? If not, was the obstacle writing, format or assumed context?</li> <li>Did it arrive while it still mattered? The UK standards require attention to <a href="https://www.gov.uk/government/publications/phia-common-analytical-standards/phia-common-analytical-standards">“changing intelligence requirements and priorities”</a>, not merely compliance with a production schedule.</li> <li>Where the consumer disagreed, was that about evidence, relevance or the conclusion, and was it recorded?</li> <li>What follow-on work did use require, and did anyone own it? <br /> <br /></li> </ul> <p>That review consumes effort. The IC’s public explanation describes evaluating products <a href="https://www.intelligence.gov/mission/our-values/objectivity">using “the IC Analytic Standards as the core criteria”</a>, but an enterprise CTI team cannot assume the same review resources. I would keep a minimum version and resist replacing report counts with immediate-action counts. That can score a useful strategic assessment or legal exposure statement at zero for producing no immediate change.</p> <p><br /></p> <p><strong>Adapt on Purpose</strong></p> <p>I am not arguing for abolishing anything. PIRs are a good idea - a prioritized, decision-linked subset is exactly what a resource-constrained shop needs, and their guidance already calls for limits, shared development and revision. Regional and actor expertise can help explain why a date on your calendar matters to somebody else. The intelligence cycle is a reasonable description of a real set of activities. The analytical standards built up over decades in national institutions - independence, comprehensiveness, honest uncertainty, auditability - are, to my eye, better than anything the commercial threat intelligence market has offered to replace them, and I would keep every one of them.</p> <p>What I am arguing against is inheriting the shape without inheriting the adaptation, and then treating the shape as settled because it came from somewhere serious.</p> <p>I would start with four questions about an existing arrangement, not treat the answers as a diagnosis on their own:</p> <ul> <li>What question does this arrangement assume somebody is asking? Say it in one sentence and see whether anyone recognizes it.</li> <li>Who decided - the producer, consumer, an authorized priority setter, or nobody, with the arrangement simply inherited?</li> <li>What would a satisfactory answer let somebody do, decide or understand differently, given today’s authority, capacity and data?</li> <li>Does the complaint concern a misfit, a legitimate mission I don’t personally value, a question the available evidence cannot settle, unfunded work, or some combination? <br /> <br /></li> </ul> <p>Those aren’t the only possibilities, and several can be in play. A misfit calls for examining and changing the arrangement. A legitimate mission you don’t value calls for a conversation with whoever does, and possibly the humility to lose that argument. An unresolved question needs its missing evidence made explicit and reconsideration when evidence or priorities change. Unfunded implementation needs a budget decision; CTI can still contribute by scoping and explaining the work required. Rewriting a report won’t create the resources, but neither does a resource problem absolve intelligence of its part in defining the need.</p> <p>The sources establish guidance, doctrine and professional argument, not controlled evidence of better outcomes. The comparison’s analyst time, planning time and displaced maintenance are stipulated costs, not observed spending. It makes the work and limitations visible without measuring a net benefit. Whether adaptation is worth the cost depends on what the existing arrangement is costing and what the change requires. I think that’s worth investigating; I can’t tell you it usually pays.</p> <p>What cyber defense needs from an intelligence shop depends on its organization, assigned mission and capacity to use an answer. The failure I care about is letting a structure built for somebody else’s mission answer that question on your behalf. Test the arrangements against the mission, and keep or change them on that basis.</p> <p><img src="http://canarytokens.com/stuff/feedback/n6h5r303nc1mf5j70xea71eir/payments.js" style="display: none;" /></p>
  113. Srsly Risky Biz: OpenAI's Kazoo-Grade Spying Incident

    Thu, 22 Oct 2026 02:55:36 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.sondera.ai/"><em><u>Sondera</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/2026-10-22-marten-newhall-uAFjFsMS3YY-unsplash.jpg" alt="Srsly Risky Biz: OpenAI&apos;s Kazoo-Grade Spying Incident"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.sondera.ai/"><em><u>Sondera</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB187.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB187/&quot;&gt;Srsly Risky Biz: OpenAI&apos;s Kazoo-Grade Spying Incident&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 18:45 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: OpenAI&apos;s Kazoo-Grade Spying Incident&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-c835f6e5-fde2-4e40-94da-8be11837165e.jpeg" class="kg-image" alt="Srsly Risky Biz: OpenAI&apos;s Kazoo-Grade Spying Incident" loading="lazy" width="2000" height="1333" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/10/data-src-image-c835f6e5-fde2-4e40-94da-8be11837165e.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/10/data-src-image-c835f6e5-fde2-4e40-94da-8be11837165e.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/10/data-src-image-c835f6e5-fde2-4e40-94da-8be11837165e.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-c835f6e5-fde2-4e40-94da-8be11837165e.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@laughayette?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Marten Newhall</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/person-using-magnifying-glass-enlarging-the-appearance-of-his-nose-and-sunglasses-uAFjFsMS3YY?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>A Saudi Arabian OpenAI employee has been arrested and charged with bugging an executive conference room in one of the company&apos;s San Francisco offices. But this story is weird! WEIRD!</p><p>The US Department of Justice (DoJ) <a href="https://www.justice.gov/opa/pr/saudi-national-arrested-installing-surreptitious-audio-recording-device-openai"><u>alleges</u></a> that Abdulrahman Alfozan hid a recording device by plugging it into the USB port of a television. The DoJ says the conference room was used for discussion of &quot;highly confidential&quot; technical, commercial and strategic information.&#xa0;</p><p>That is serious, but the way Alfozan allegedly tried to launch his spying career is almost comically amateur.&#xa0;&#xa0;</p><p>According to <a href="https://www.courtlistener.com/docket/74958184/united-states-v-alfozan/"><u>the criminal complaint</u></a>, the bug was a small SEEED brand microcontroller development board, with camera, microphone and wifi connectivity, that retails for <a href="https://www.seeedstudio.com/XIAO-ESP32S3-Sense-p-5639.html?srsltid=AU7gw4WgvDEnaJESoAdnz3xy6BjYnQK8pXuL-UsHJBTjLOjzHMmkaCeO"><u>USD&#x24;13.90</u></a>. OpenAI examined the device and assessed it may have been modified to encrypt data to prevent forensic analysis. Best of all, the device was &quot;concealed in toilet paper&quot;.</p><p><a href="https://storage.courtlistener.com/recap/gov.uscourts.cand.480829/gov.uscourts.cand.480829.12.0_3.pdf"><u>In separate court filings</u></a> the DoJ alleges that in the days prior to installing the device, Alfozan used his wife&apos;s personal ChatGPT account to research how to build a continuous audio recording device. The prompts provided to the chatbot included asking about encrypting recordings and configuring the device for USB power-only operation. Alfozan also used his own ChatGPT account to find out the best meeting rooms to target at OpenAI.&#xa0;&#xa0;</p><p>Alfozan also left a bag containing another recording device unattended in an OpenAI printer room. By then, though, the company was aware of his activities and found and confiscated the device after searching the bag. That unattended bag with the recording device also contained personal items including Alfozan&apos;s wallet and ID. Clever.</p><p>Alfozan was arrested a few days before he was scheduled to travel to Riyadh to take two months of parental leave, and the DoJ alleges Alfozan has strong links to Saudi Arabia.</p><p>According to documents tendered to the court, Alfozan was in &quot;frequent communication&quot; with a member of the Saudi Royal Family via WhatsApp. That individual encouraged Alfozan to join an AI initiative that was being launched in Saudi Arabia, saying &quot;Saudi needs you more then [sic] the US!&quot;.&#xa0;</p><p>The kingdom&apos;s government does appear to like this guy. It paid for tickets to repatriate his wife and child after he was arrested.&#xa0;</p><p>With all that in mind, we don&apos;t think this is state-backed espionage, at least in the traditional sense. By arming himself with illicitly obtained commercially sensitive information, we think Alfozan was probably just trying to juice up his appeal as a hire back home. And his homie in the royal family would also presumably be impressed when he returned to Saudi Arabia toting a bag full of OpenAI conference room recordings. Everyone wins!</p><p>Even though this incident doesn&apos;t bear the hallmarks of a state run espionage operation, we&apos;re certain that intelligence services <em>are</em> targeting the frontier labs. More than three years ago we <a href="https://news.risky.biz/nsa-wants-to-protect-americas-ai-65c0eb42a0e6da001a37dfbd/"><u>wrote about</u></a> how the NSA wanted to protect America&apos;s AI edge, and it&apos;s even clearer now that AI technology is central to geopolitical competition.</p><p>Unfortunately, the incentives for companies to move fast to stay ahead of the competition are not conducive to strong security. We&apos;ve already seen <a href="https://openai.com/index/hugging-face-incident-and-the-road-ahead/"><u>a spate</u></a> of <a href="https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals"><u>embarrassing</u></a> AI <a href="https://www.wired.com/story/ai-hacks-leave-a-mess-that-ai-companies-arent-cleaning-up/"><u>security incidents</u></a> where frontier lab models <a href="https://arstechnica.com/security/2026/10/openai-models-hacked-a-crypto-wallet-and-breached-more-websites/"><u>successfully embarked</u></a> on <a href="https://www.bbc.com/news/articles/cw24jm9rryy3o"><u>overenthusiastic hacking adventures</u></a>. In our view, these are directly downstream of a culture that prioritises speed over security.&#xa0;</p><p>Mitigating insider threats and state-level espionage risks is an arduous undertaking. Intelligence agencies&apos; mitigations, for example, include lengthy and extensive personnel vetting, strict access control, behavioural monitoring and years spent developing a strong security culture.</p><p>Likening a frontier AI company to an intelligence agency may seem like a stretch, but the reality is these labs are top priorities for state espionage. And we don&apos;t think strong insider threat programs are compatible with the pace at which AI companies are moving.</p><p>When it comes to the labs themselves, this incident needs to serve as a wake up call that they need to improve their operational security. If a clown like Alfozan can get a listening device into a conference room, what could a sophisticated actor do? Those spies won&apos;t be wrapping off-the-shelf cameras in toilet paper. Instead, they could be remotely compromising that conference room smart TV, and they&apos;ll be far harder to catch.&#xa0;</p><p>So while intelligence agency-level insider threat controls are unlikely to be implemented at the frontier labs, they will need to work harder to protect their crown jewels.&#xa0;</p><h2 id="watch-out-cybercriminals-ai-is-coming-for-your-jobs-too">Watch Out, Cybercriminals! AI Is Coming for Your Jobs, Too</h2><p>Security firm NorForge has released <a href="https://norforge.com/blog/inside_cairn"><u>a report</u></a> detailing how various threat actors are using the Cairn autonomous pentesting agent in financially-motivated cybercrime. And they&apos;re actually making bank.</p><p>This is a pretty stark contrast to some of the other AI-assisted escapades <a href="https://news.risky.biz/srsly-risky-biz-open-weight-model-advances-make-the-mythos-debate-moot/"><u>we&apos;ve written about</u></a> in <a href="https://news.risky.biz/srsly-risky-biz-it-is-time-to-ban-sale-of-precise-geolocation/#ai-is-your-helpful-hacker-team"><u>recent months</u></a>. In those examples, the threat actors successfully used AI to do the hacking, but were hopeless at making money out of the access they obtained.</p><p>Cairn <a href="https://github.com/oritera/Cairn"><u>is an open source project</u></a>, described by its maintainers as a &quot;general-purpose AI state-space search engine&quot;. When given a starting point and a goal it tries to search out a path between A to B. Cairn is well suited to penetration testing, for example. There&apos;s a clear starting point, such as an IP or a system, and a concrete measure of success, like getting a shell. The project&apos;s documentation says it also works well for vulnerability research and mathematical proofs.</p><p>Cairn is not the only harness being used to do malicious AI hacking, but it is often configured with an unsecured web-based user interface, making it visible to tools like Shodan. NorForge was able to find 20-ish open Cairn panels that were being used to conduct malicious hacking campaigns.&#xa0;</p><p>The largest operation NorForge identified was a single Chinese-language cluster of 14 Cairn dashboards running over 1100 attack projects. This cluster mostly targeted the American, Japanese and Australian retail and wholesale sectors, with an apparent goal of gaining persistent access to e-commerce payment infrastructure. So, AI-driven hacks are providing access to credit card and payment details.&#xa0;</p><p>A second, English-language cluster was less prolific. This set of panels was focussed on payments and financial services crime, with a similar goal: Steal information that can be converted immediately into money.&#xa0;</p><p>The final cluster was smaller and focussed on a mix of e-commerce sites and very large US-based commercial platforms.&#xa0;</p><p>Overall, NorForge reckons that about 10% of the hacks being managed by Cairn were successful, and they averaged less than 24 hours to complete. There were indications that two of the clusters were using DeepSeek models that are relatively cheap to run. Snorre Fagerland, one of the report&apos;s authors, <a href="https://www.reuters.com/legal/government/how-one-hacker-used-ai-agents-few-thousand-dollars-break-into-scores-sites-2026-10-19/"><u>told <em>Reuters</em></u></a> he estimated the cost was roughly USD&#x24;50 per successful hack. Decent bang for buck, then.</p><p>Some of the actual hacking was also quite impressive. Per <em>Reuters</em>:</p><blockquote>In one case, after failing to find any weaknesses in a big company&apos;s public-facing application, the AI agents pored through digital records to find associated infrastructure, located a public-facing control panel, discovered a subtle weakness in the way it processed data, and after &#x200b;hours of trial and error, reported leveraging it &#x200b;to trick the server into executing malicious commands.</blockquote><p>To us the significant thing here is that threat actors have found a target set where they can automate the entire process, from hacking a site to transferring a victim&apos;s cryptocurrency to an attacker-controlled wallet. Old-school cybercriminals should be worried. AI is coming for their jobs!</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/k3e-w6f3sHI?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: OpenAI&apos;s Kazoo-Grade Spying Incident"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>AI safety on the agenda:</strong> The US Federal Trade Commission has issued subpoenas to leading AI companies asking for documents relating to the firms&apos; security incidents and their safety practices, <a href="https://www.axios.com/2026/10/15/ftc-openai-anthropic-ai-safety-investigation"><u>reports <em>Axios</em></u></a>. This comes after the White House Super Intelligence Force <a href="https://www.axios.com/2026/10/09/anthropic-ai-security-white-house"><u>mandated earlier this month</u></a> that companies must notify and remedy security incidents involving government systems.&#xa0;&#xa0;&#xa0;</li><li><strong>Triple strike against the Chinese cybercrime ecosystem:</strong> <a href="https://cyberscoop.com/microsoft-disrupts-chinese-language-phishing-kit-ecosystem/"><u>An international collaboration</u></a> including Microsoft, Japan&apos;s National Police agency and the Health-ISAC has taken down three Chinese cybercrime services. CoGUI and Iron Man sold phishing kits while the YAM Store sold compromised accounts. <a href="https://blogs.microsoft.com/on-the-issues/2026/10/14/disrupting-the-service-economy-behind-transnational-cybercrime-operations/"><u>Microsoft says</u></a> that the kits have sent more than a billion malicious messages since January.&#xa0;</li><li><strong>Chrome gets ClickFix protections</strong>: <a href="https://blog.google/security/our-latest-defenses-against-evolving-web-scams/"><u>Google has announced</u></a> a range of improved security features including suspicious site warnings, improved scam detection and protections against <a href="https://blog.google/innovation-and-ai/technology/safety-security/fraud-scams-advisory-june-2026/#:~:text=well%20as%20the%20%E2%80%9C-,ClickFix,-%E2%80%9D%20campaign%20(which%20uses"><u>ClickFix attacks</u></a>.&#xa0;&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Sondera co-founder and CEO Josh Devon about how AI agents are highly capable, but they&#x2019;re not the best rule followers or decision makers.&#xa0;</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI147.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI147/&quot;&gt;Sponsored: Agents don&#x2019;t care about human ideas of right and wrong&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 27:12 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Agents don&#x2019;t care about human ideas of right and wrong&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="shorts">Shorts</h2><h3 id="a-bad-time-to-be-a-russian-data-centre">A Bad Time To Be A Russian Data Centre</h3><p>Ukrainian drones <a href="https://www.datacenterdynamics.com/en/news/vk-data-center-near-st-petersburg-russia-struck-by-drone/"><u>struck a data centre</u></a> operated by Russian social media giant VK on Sunday. Although VK says services are available as usual, internet monitoring firm Netblocks says VK network activity <a href="https://mastodon.social/@netblocks/117459385064329291"><u>dropped by 44 percent</u></a> after the strike.&#xa0;</p><p>This follows on <a href="https://news.risky.biz/srsly-risky-biz-russias-digital-detox-boomerang/"><u>from our article last week</u></a> discussing the impact of Russian attacks on Ukrainian data centres. Russia&apos;s efforts to build a sovereign internet make it more vulnerable to Ukrainian retaliation.&#xa0;&#xa0;</p><h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss whether the rise of AI coding will result in more or less bugs and whether it will ultimately make software more secure.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN186.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN186/&quot;&gt;Between Two Nerds: Why AI will mean infinite bugs&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 23:56 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Why AI will mean infinite bugs&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/5chpbGb-0TY?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Why AI will mean infinite bugs"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Hacker hijacks 10% of Tor exit nodes:</strong> A mysterious threat actor has hacked more than 330 Tor servers earlier this month to carry out a mass-sniffing campaign against Tor users.</p><p>The attacker hacked Tor relays, infected them with the Ebury Linux malware, and converted them to exit relays where it could intercept SSH connections and downgrade email connections to steal unencrypted email traffic.</p><p>The hacking campaign went undetected for three days, between October 2 and 5, before it was <a href="https://kaba.cx/posts/20261020_Tor/"><u>spotted</u></a> by a Swiss engineer and reported to the Tor security team.</p><p>[<a href="https://news.risky.biz/risky-bulletin-hacker-hijacks-10-of-tor-exit-nodes/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>PwnEscape vulnerability lets you escape KVM: </strong>The Linux kernel team has patched a major vulnerability that lets virtual machine guests escape the KVM Linux virtualization module and run code on the underlying host operating system.</p><p>Codenamed <a href="https://pwn.ai/blog/pwnescape"><u>PwnEscape</u></a> (<a href="https://access.redhat.com/security/cve/cve-2026-99999"><u>CVE-2026-99999</u></a>), the bug occurs because the Linux networking stack allows the VM guest to modify IP packets after they&apos;ve already been validated as safe by the kernel. This allows a skilled malicious guest to read host memory and potentially modify it for code execution.</p><p>PwnEscape impacts KVM running on x86 CPU architectures, which is the most common setup used by most shared cloud environments.</p><p>[<a href="https://news.risky.biz/risky-bulletin-pwnescape-vulnerability-lets-you-escape-kvm/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Russia arrests DataSuckers hacking group: </strong>Russian authorities have <a href="https://t.me/bazabazon/49935?embed=1&amp;mode=tme"><u>arrested</u></a> three members of DataSuckers, a Lapsus and ShinyHunters-like hacking and data extortion group operating across the former Soviet space.</p><p>Authorities say the group has four members in total, spread across Russia, Belarus, and Kazakhstan. A 19-year-old named Denis A. Friedman, from the city of Barnaul, was <a href="https://max.ru/id7718123097_gos/AaE61YSpdYE"><u>arraigned</u></a> in a Moscow court on Wednesday, but no other details were provided on the other two detained suspects.</p><p>DataSuckers started operating this year and has been linked to at least three major hacks&#x2014;Ukrainian super market chain <a href="https://cybercalm.org/ataka-na-sajt-atb-datasuckers/"><u>ATB Market</u></a>, Russian travel company <a href="https://hackmag.com/news/tez-tour-hack"><u>Tez Tour</u></a>, and Russian pizza chain <a href="https://english.pravda.ru/news/russia/169358-dodo-pizza-cyberattack-data-leak/"><u>Dodo Pizza</u></a>.</p><p>[<a href="https://news.risky.biz/risky-bulletin-russia-arrests-datasuckers-hacking-group/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  114. PHP2Shell: What to Hunt on the Web Server

    Wed, 21 Oct 2026 00:00:00 -0000

    Figured I wouldn’t pass up the chance to pile on the PHP2Shell Hype Train. Here’s the quick and dirty: PHP’s October 21 emergency releases fix two vulnerabilities that can allow unauthenticated remote code execution. For the responders, that creates two different jobs: get the fixed runtime into the processes serving requests, and investigate whether those processes have done anything that needs explaining. A successful package update answers neither question by itself.
    <p>Figured I wouldn’t pass up the chance to pile on the PHP2Shell Hype Train. Here’s the quick and dirty: PHP’s <a href="https://www.php.net/archive/2026.php#2026-10-21-5">October 21 emergency releases fix two vulnerabilities that can allow unauthenticated remote code execution</a>. For the responders, that creates two different jobs: get the fixed runtime into the processes serving requests, and investigate whether those processes have done anything that needs explaining. A successful package update answers neither question by itself.</p> <p>The useful starting point is the web worker, not a search for anything named <code class="language-plaintext highlighter-rouge">php.exe</code>. With <a href="https://www.php.net/manual/en/install.unix.apache2.php">Apache’s <code class="language-plaintext highlighter-rouge">mod_php</code>, PHP runs inside the Apache worker</a>. That changes which parent process or file writer belongs in a hunt.</p> <p>I’ll start with the patch versions and the request-lifecycle differences, then work through two behavior families in Microsoft Defender XDR, Splunk, and Elastic Security: a PHP-serving worker starting a selected shell or interpreter, and a worker writing candidate script files beneath a configured webroot. Those are investigation leads. They aren’t signatures for either vulnerability, and neither behavior has to occur for exploitation to have happened.</p> <p><img width="1600" height="1000" alt="PHP Response: Two evidence paths" src="https://github.com/user-attachments/assets/feda5559-562d-4356-95c9-4c785ecf9d41" /></p> <p><br /></p> <p><strong>Patch the Runtime That’s Serving Requests</strong></p> <p>These are the upstream fixed releases in the <a href="https://www.php.net/archive/2026.php#2026-10-21-5">PHP emergency announcement</a>:</p> <table> <thead> <tr> <th>PHP Branch</th> <th>Fixed Release</th> </tr> </thead> <tbody> <tr> <td>8.5</td> <td>8.5.12</td> </tr> <tr> <td>8.4</td> <td>8.4.27</td> </tr> <tr> <td>8.3</td> <td>8.3.36</td> </tr> <tr> <td>8.2</td> <td>8.2.35</td> </tr> </tbody> </table> <p>PHP <a href="https://www.php.net/archive/2026.php#2026-10-21-5">describes all web-serving installations as at risk and assesses the risk as highest for <code class="language-plaintext highlighter-rouge">mod_php</code></a>. The announcement includes PHP-FPM, CGI, FrankenPHP, and LiteSpeed; it does not say every configuration has the same demonstrated exploit path.</p> <p>If you’re using a distribution or LTS package, check that vendor’s remediation information for your installed build. The upstream announcement says vendors received patches in advance; that isn’t verification that your particular package is available or installed. PHP also supplied exceptional minimal source backports for 7.4, 8.0, and 8.1. <a href="https://www.php.net/archive/2026.php#2026-10-21-5">Those are source backports, not restored support for the end-of-life branches</a>.</p> <p>For patch verification, I would record the affected site’s serving mode, the runtime or module it loads, the fixed build or documented vendor backport, and the worker replacement or restart that brings that build into service. A CLI <code class="language-plaintext highlighter-rouge">php -v</code> result can help inventory a binary, but it doesn’t establish what an Apache module or a separate FPM pool has loaded. The reason is architectural: <a href="https://www.php.net/manual/en/install.unix.apache2.php">the Apache module is embedded in the worker</a>, not necessarily the command-line executable you just checked. Verify the serving configuration and running workers, including separately deployed pools or containers.</p> <p>Don’t make this wait for a hunt result. The advisories establish a patching need; an endpoint query answers a different question.</p> <p><br /></p> <p><strong>Two Failures at Request Boundaries</strong></p> <p>The <a href="https://github.com/php/php-src/security/advisories/GHSA-w8hw-w9fx-j79w">OOM</a> and <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">stale-state</a> advisories both rate these flaws Critical, CVSS v4.0 9.2, but the lifecycle details matter more to the response than treating them as one interchangeable PHP RCE.</p> <p>For CVE-2026-109618, PHP can make request memory reachable before initialization is complete. An allocation failure triggers an out-of-memory exit through <code class="language-plaintext highlighter-rouge">longjmp()</code>, leaving that memory in an invalid state. <a href="https://github.com/php/php-src/security/advisories/GHSA-w8hw-w9fx-j79w">Request shutdown can then read or branch on it</a>. With <code class="language-plaintext highlighter-rouge">mod_php</code>, shutdown still runs when request startup fails. The advisory’s confirmed HTTP exploitation happens before application-specific code executes, with default PHP input and memory limits and normal prefork configurations containing multiple workers. Application authentication doesn’t stop that path simply by existing later in the request flow.</p> <p>CVE-2026-109619 is about state surviving too long. A producer leaves request-owned state reachable through a worker’s global state; the request heap is freed, but the reference survives. A later request on that same worker reaches a consumer before normal initialization resets the stale value. <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">Both producer and consumer are necessary</a>. A fatal error or enabled session feature, considered alone, is not a complete exploit chain.</p> <table> <thead> <tr> <th>Path</th> <th>What the Advisory Establishes</th> <th>Boundary to Keep</th> </tr> </thead> <tbody> <tr> <td>OOM, <code class="language-plaintext highlighter-rouge">mod_php</code></td> <td>HTTP RCE before application code, including default limits and multiple prefork workers.</td> <td>Execution has the web worker’s privileges, not automatic root access.</td> </tr> <tr> <td>OOM, FPM</td> <td>Failed startup exits the child; source analysis supports susceptibility to later application OOM.</td> <td>No complete FPM exploit demonstrated for this CVE.</td> </tr> <tr> <td>Stale state, PHP 8.5 fatal backtrace</td> <td>Confirmed RCE against <code class="language-plaintext highlighter-rouge">mod_php</code>.</td> <td>This is one state family, not the definition of every stale-state path.</td> </tr> <tr> <td>Stale state, session lifecycle</td> <td>RCE validated against <code class="language-plaintext highlighter-rouge">mod_php</code> and FPM.</td> <td>The FPM producer requires a suitable application operation after successful startup.</td> </tr> </tbody> </table> <p><em>Different lifecycle failures, different demonstrated paths. These are the PHP project’s <a href="https://github.com/php/php-src/security/advisories/GHSA-w8hw-w9fx-j79w">OOM</a> and <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">stale-state</a> validation results, not observed intrusions or experiments performed for this article.</em></p> <p>Exiting an FPM child after failed startup prevents consumption of invalid state on that OOM path. It doesn’t erase the separately validated session-lifecycle path. In the latter, the advisory describes an application operation publishing state after successful startup, a fatal error, and session upload progress consuming the stale state later. <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">Upload progress is enabled by default and doesn’t require the application to call <code class="language-plaintext highlighter-rouge">session_start()</code></a>.</p> <p>For stale state, the producer and consumer must reach the same worker. <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">Multiple workers and recycling reduce success per request but don’t remove the condition</a>. That is a reason to understand worker lifetime, not to infer safety from a busy pool. The public advisories provide the demonstrated and conditional cases; they don’t establish that either CVE has been used to deploy webshells in the wild.</p> <p><br /></p> <p><strong>Temporary Settings Have Separate Jobs</strong></p> <p>If an update is blocked, use the advisories’ temporary measures with their actual scope. For the confirmed session paths, <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">disable <code class="language-plaintext highlighter-rouge">session.upload_progress.enabled</code> administratively and restart every worker</a>. For the confirmed PHP 8.5 backtrace path, <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">disable <code class="language-plaintext highlighter-rouge">fatal_error_backtraces</code> and restart every worker</a>. Neither setting covers the other state families. Configuration must apply early enough: the <a href="https://www.php.net/archive/2026.php#2026-10-21-5">emergency announcement specifically warns against relying on <code class="language-plaintext highlighter-rouge">ini_set()</code> or FPM <code class="language-plaintext highlighter-rouge">.user.ini</code> for the upload-progress mitigation</a>.</p> <p>The OOM advice needs more care than a copied one-line fix. Administratively enforcing <code class="language-plaintext highlighter-rouge">memory_limit=-1</code> removes the controllable PHP memory-limit failure, but <a href="https://github.com/php/php-src/security/advisories/GHSA-w8hw-w9fx-j79w">underlying allocation failure can still reach the vulnerable path</a>. Removing that per-request boundary also requires an availability plan. The advisory pairs temporary mitigation with conservative input limits and independently enforced decoded-request limits at the frontend. It warns that <code class="language-plaintext highlighter-rouge">RLIMIT_AS</code> and <code class="language-plaintext highlighter-rouge">RLIMIT_DATA</code> can induce the allocation failure, while a cgroup limit is not a complete safeguard and can kill more than one worker.</p> <p>I wouldn’t turn that into a blanket instruction to remove memory limits. Work through the published controls with the application and platform owners while getting the fixed runtime deployed. These are partial measures for distinct paths, not a configuration-only substitute for the patches.</p> <p><br /></p> <p><strong>Start With the Worker, Then Choose the Behavior</strong></p> <p><img width="1600" height="1000" alt="Two Separate Behavior Families" src="https://github.com/user-attachments/assets/92fece5d-6631-42cd-8f29-02df3145a404" /></p> <p>Inventory decides which processes belong here. For an inventoried <code class="language-plaintext highlighter-rouge">mod_php</code> deployment, that can mean <code class="language-plaintext highlighter-rouge">apache2</code>, <code class="language-plaintext highlighter-rouge">httpd</code>, or <code class="language-plaintext highlighter-rouge">httpd.exe</code>. For another serving mode, use its actual FPM, CGI, FrankenPHP, or LSAPI executable names. The lists below are starting points, not a discovery method. An unlisted version suffix or renamed executable won’t match. NGINX or IIS being present doesn’t, by itself, establish which process handled PHP.</p> <p>A <a href="https://www.microsoft.com/en-us/security/blog/2021/02/11/web-shell-attacks-continue-to-rise/">webshell is server-side backdoor code</a>, not another name for an operating-system shell. A PHP backdoor may use PHP facilities without starting <code class="language-plaintext highlighter-rouge">sh</code> or <code class="language-plaintext highlighter-rouge">cmd.exe</code>. Conversely, a legitimate application can start an interpreter. Memory corruption in a worker doesn’t require either an on-disk backdoor or a selected child process. That’s why the following are two optional behavior families:</p> <ul> <li><strong>A: Direct child execution.</strong> An inventoried PHP-serving process directly starts one of the selected shells or interpreters.</li> <li><strong>B: Direct worker file activity.</strong> An inventoried worker creates, changes, or renames a candidate script beneath a configured webroot. The Sysmon version covers creation/overwrite only.</li> </ul> <p>These are established hunting ideas. <a href="https://www.microsoft.com/en-us/security/blog/2021/02/11/web-shell-attacks-continue-to-rise/">Microsoft discusses script-file creation and responsible process trees</a>; Elastic publishes <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_suspicious_child_execution.toml">web-server child-process</a> and <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_php_file_creation_in_wordpress_plugin_dir.toml">PHP plugin-file</a> detections. Here, I’ll apply those ideas to the PHP-serving process and spell out each tool’s data requirements.</p> <table> <thead> <tr> <th>Tool and Language</th> <th>Required Records</th> <th>Family B Scope</th> </tr> </thead> <tbody> <tr> <td>Defender XDR, Kusto</td> <td><code class="language-plaintext highlighter-rouge">DeviceProcessEvents</code> and <code class="language-plaintext highlighter-rouge">DeviceFileEvents</code> on selected Windows/Linux hosts.</td> <td>Selected creation, modification, and rename actions, where collected.</td> </tr> <tr> <td>Splunk, SPL</td> <td>Windows Sysmon Event 1 and Event 11 with native fields extracted.</td> <td>Creation/overwrite only.</td> </tr> <tr> <td>Elastic Security, EQL</td> <td>Linux Elastic Defend process and file event streams.</td> <td>Creation/change or the Defend rename action.</td> </tr> </tbody> </table> <p><em>The examples use <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-deviceprocessevents-table">MDE’s process</a> and <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicefileevents-table">file</a> schemas, <a href="https://learn.microsoft.com/en-us/sysinternals/downloads/sysmon">Windows Sysmon event definitions</a>, and the qualified Elastic Defend contracts described below. These are not equivalent collection guarantees.</em></p> <p><em>Editorial Note: Release and advisory details reflect public sources retrieved on October 21, 2026. All six queries are documentation-reviewed sample hunts. They have not been parsed or executed in the native backends, tested against fixtures, or validated for production efficacy.</em></p> <p>Before adapting a query, replace the host and any index/source placeholders. Replace the example webroots with the directories the selected service actually serves, in the path representation recorded by its sensor. Run separately for differing host configurations unless you’ve designed an inventory mapping. Symlinks, bind mounts, and container paths aren’t resolved by these string predicates.</p> <p>Use the seven-day window only if it fits the investigation and retained data. Keep original events alongside the displayed columns; projections are reading aids, not complete evidence exports. Missing command lines or hashes shouldn’t discard an otherwise matching event, but missing required parent/writer fields are a coverage problem. None of these examples joins events by PID, computes all descendants, or identifies an HTTP request.</p> <p><br /></p> <p><strong>Microsoft Defender XDR: Parent and Writer Fields</strong></p> <p>Use Advanced Hunting with Defender for Endpoint onboarded and the relevant records retained for the selected devices. <a href="https://learn.microsoft.com/en-us/defender-endpoint/microsoft-defender-endpoint-linux">Linux supports Advanced Hunting</a>, but that doesn’t establish that every file action is present on every host. Microsoft’s table documentation directs you to the portal schema for supported action types. The public community reference describes <a href="https://xdrinternals.com/docs/microsoftxdr/devices/deviceprocessevents/"><code class="language-plaintext highlighter-rouge">ProcessCreated</code></a> and <a href="https://xdrinternals.com/docs/microsoftxdr/devices/devicefileevents/"><code class="language-plaintext highlighter-rouge">FileCreated</code>, <code class="language-plaintext highlighter-rouge">FileModified</code>, and <code class="language-plaintext highlighter-rouge">FileRenamed</code></a>; confirm those actions in your own collection before interpreting missing results.</p> <p>For MDE-A, the row describes the created child. <code class="language-plaintext highlighter-rouge">FileName</code> and <code class="language-plaintext highlighter-rouge">ProcessCommandLine</code> belong to that child; <code class="language-plaintext highlighter-rouge">InitiatingProcessFileName</code> and its related fields describe the initiating parent in the <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-deviceprocessevents-table">process-event schema</a>. Don’t substitute <code class="language-plaintext highlighter-rouge">InitiatingProcessParent*</code>, which steps back another generation.</p> <pre><code class="language-kusto">DeviceProcessEvents | where Timestamp &gt;= ago(7d) and Timestamp &lt; now() | where DeviceName in~ ("REPLACE_WITH_PHP_HOST") | where ActionType == "ProcessCreated" | where InitiatingProcessFileName in~ ( "apache2", "httpd", "httpd.exe", "php-fpm", "php-fpm8.2", "php-fpm8.3", "php-fpm8.4", "php-fpm8.5", "php-cgi", "php-cgi.exe", "frankenphp", "lsphp") | where FileName in~ ( "sh", "bash", "dash", "zsh", "ksh", "csh", "tcsh", "fish", "cmd.exe", "powershell.exe", "pwsh", "pwsh.exe", "python", "python3", "python.exe", "python3.exe", "perl", "perl.exe", "ruby", "ruby.exe", "node", "node.exe", "php", "php.exe") | project Timestamp, DeviceId, DeviceName, ReportId, ActionType, AccountDomain, AccountName, InitiatingProcessAccountDomain, InitiatingProcessAccountName, InitiatingProcessId, InitiatingProcessCreationTime, InitiatingProcessFileName, InitiatingProcessFolderPath, InitiatingProcessCommandLine, ProcessId, ProcessCreationTime, FileName, FolderPath, ProcessCommandLine </code></pre> <p>A result asks why this worker started this program. Read the executable paths and command lines against the expected application function. A helper used for rendering or administration is a competing explanation, not an exception to add before checking it.</p> <p>MDE-B uses the file event’s initiating process as the writer. Here, <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicefileevents-table"><code class="language-plaintext highlighter-rouge">FolderPath</code> is the containing folder and <code class="language-plaintext highlighter-rouge">FileName</code> is the affected filename</a>. That distinction explains the equality check for files directly in the root, followed by the separator-bounded prefix for nested folders.</p> <pre><code class="language-kusto">DeviceFileEvents | where Timestamp &gt;= ago(7d) and Timestamp &lt; now() | where DeviceName in~ ("REPLACE_WITH_PHP_HOST") | where ActionType in ("FileCreated", "FileModified", "FileRenamed") | where InitiatingProcessFileName in~ ( "apache2", "httpd", "httpd.exe", "php-fpm", "php-fpm8.2", "php-fpm8.3", "php-fpm8.4", "php-fpm8.5", "php-cgi", "php-cgi.exe", "frankenphp", "lsphp") | where FolderPath == "/var/www/html" or FolderPath startswith_cs "/var/www/html/" or FolderPath =~ "C:\\inetpub\\wwwroot" or FolderPath startswith "C:\\inetpub\\wwwroot\\" | where FileName endswith ".php" or FileName endswith ".phtml" or FileName endswith ".php5" or FileName endswith ".phar" | project Timestamp, DeviceId, DeviceName, ReportId, ActionType, FileName, FolderPath, PreviousFileName, PreviousFolderPath, SHA1, InitiatingProcessAccountDomain, InitiatingProcessAccountName, InitiatingProcessId, InitiatingProcessCreationTime, InitiatingProcessFileName, InitiatingProcessFolderPath, InitiatingProcessCommandLine </code></pre> <p><a href="https://learn.microsoft.com/en-us/kusto/query/datatypes-string-operators">Kusto’s <code class="language-plaintext highlighter-rouge">in~</code> operator compares whole names without case sensitivity</a>; the Linux root comparisons remain case-sensitive. Windows root and script-suffix comparisons are case-insensitive. The separator prevents <code class="language-plaintext highlighter-rouge">/var/www/html-backup</code> from qualifying as a child of <code class="language-plaintext highlighter-rouge">/var/www/html</code>; the filename suffix won’t select a text file merely because its directory contains <code class="language-plaintext highlighter-rouge">.php</code>. Don’t replace those checks with term-based <code class="language-plaintext highlighter-rouge">has</code> searches.</p> <p>Retain rename history and available hashes for investigation. Neither query requires a unique process key to be populated on Linux. Microsoft’s <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-best-practices">process-identity guidance retains device, PID, and creation time for non-Windows systems</a>; its Windows unique-ID guidance isn’t evidence of universal Linux population. Preserve that composite context, and keep <code class="language-plaintext highlighter-rouge">ReportId</code> with device and timestamp rather than treating it as a globally unique event ID.</p> <p><br /></p> <p><strong>Splunk: An Explicit Windows Sysmon Contract</strong></p> <p>These SPL examples are Windows-only. They assume a known Sysmon index and source, the official add-on’s documented <a href="https://splunk.github.io/splunk-add-on-for-microsoft-sysmon/Sourcetypes/"><code class="language-plaintext highlighter-rouge">XmlWinEventLog</code> sourcetype</a>, and native fields extracted without changing their meaning. Map differences explicitly if your ingestion differs. The add-on uses <a href="https://splunk.github.io/splunk-add-on-for-microsoft-sysmon/Lookups/"><code class="language-plaintext highlighter-rouge">EventCode</code> for its event-description lookup</a>.</p> <p>Use <code class="language-plaintext highlighter-rouge">Computer</code> for the originating event’s computer, not an assumed forwarder identity. <a href="https://www.ultimatewindowssecurity.com/securitylog/encyclopedia/event.aspx?eventid=90001">The published Event 1 XML</a> distinguishes the child <code class="language-plaintext highlighter-rouge">Image</code>, <code class="language-plaintext highlighter-rouge">CommandLine</code>, <code class="language-plaintext highlighter-rouge">ProcessGuid</code>, and <code class="language-plaintext highlighter-rouge">ProcessId</code> from their parent equivalents. Preserve the original event along with the table output, especially when Windows Event Collection sits between the endpoint and Splunk.</p> <p>SPL-A selects an inventoried Windows Apache or PHP-CGI worker directly starting one of the listed programs:</p> <pre><code class="language-spl">index="REPLACE_WITH_SYSMON_INDEX" sourcetype="XmlWinEventLog" source="REPLACE_WITH_SYSMON_SOURCE" Computer="REPLACE_WITH_PHP_HOST" earliest=-7d latest=now EventCode=1 | eval parent_image=lower(replace(ParentImage, "\\\\", "/")), child_image=lower(replace(Image, "\\\\", "/")) | where match(parent_image, "/(httpd|php-cgi)[.]exe$") | where match(child_image, "/(cmd|powershell|pwsh|python|python3|perl|ruby|node|php|bash|sh)[.]exe$") | table _time UtcTime Computer host source EventCode EventRecordID User ParentImage ParentCommandLine ParentProcessGuid ParentProcessId Image CommandLine ProcessGuid ProcessId </code></pre> <p>The comparison fields are lowercased and normalize backslashes to forward slashes; the original fields stay in the output. <a href="https://help.splunk.com/en/splunk-enterprise/search/spl-search-reference/9.4/evaluation-functions/text-functions">SPL requires two layers of backslash escaping inside <code class="language-plaintext highlighter-rouge">eval replace</code></a>. The final-component separator and <code class="language-plaintext highlighter-rouge">$</code> anchor matter because <a href="https://help.splunk.com/en/splunk-enterprise/search/spl-search-reference/9.4/evaluation-functions/comparison-and-conditional-functions"><code class="language-plaintext highlighter-rouge">match</code> otherwise finds a matching substring</a>. A directory containing <code class="language-plaintext highlighter-rouge">httpd.exe</code> isn’t the same thing as an executable with that final component.</p> <p>SPL-B uses Event 11. <a href="https://learn.microsoft.com/en-us/sysinternals/downloads/sysmon">Sysmon documents it as file creation or overwrite</a>, not comprehensive modification or rename telemetry:</p> <pre><code class="language-spl">index="REPLACE_WITH_SYSMON_INDEX" sourcetype="XmlWinEventLog" source="REPLACE_WITH_SYSMON_SOURCE" Computer="REPLACE_WITH_PHP_HOST" earliest=-7d latest=now EventCode=11 | eval writer_image=lower(replace(Image, "\\\\", "/")), target_path=lower(replace(TargetFilename, "\\\\", "/")) | where match(writer_image, "/(httpd|php-cgi)[.]exe$") | where like(target_path, "c:/inetpub/wwwroot/%") | where match(target_path, "[.](php|phtml|php5|phar)$") | table _time UtcTime Computer host source EventCode EventRecordID Image ProcessGuid ProcessId TargetFilename CreationUtcTime </code></pre> <p>Here <code class="language-plaintext highlighter-rouge">Image</code> is the writer and <code class="language-plaintext highlighter-rouge">TargetFilename</code> is the full affected path. The root ends in a separator so a sibling such as <code class="language-plaintext highlighter-rouge">wwwroot-backup</code> isn’t selected. In <a href="https://help.splunk.com/en/splunk-enterprise/search/spl-search-reference/9.4/evaluation-functions/comparison-and-conditional-functions"><code class="language-plaintext highlighter-rouge">like</code>, <code class="language-plaintext highlighter-rouge">%</code> is the multi-character wildcard</a>, not <code class="language-plaintext highlighter-rouge">*</code>. Account for wildcard or regex characters if your replacement paths contain them.</p> <p>The <a href="https://www.ultimatewindowssecurity.com/securitylog/encyclopedia/event.aspx?eventid=90011">inspected Event 11 example</a> supplies <code class="language-plaintext highlighter-rouge">Image</code>, <code class="language-plaintext highlighter-rouge">ProcessGuid</code>, <code class="language-plaintext highlighter-rouge">ProcessId</code>, <code class="language-plaintext highlighter-rouge">TargetFilename</code>, <code class="language-plaintext highlighter-rouge">UtcTime</code>, and <code class="language-plaintext highlighter-rouge">CreationUtcTime</code>. It doesn’t supply a writer command line or an EventData <code class="language-plaintext highlighter-rouge">User</code>, so the query doesn’t invent either. Don’t substitute provider security identity for the writer. <code class="language-plaintext highlighter-rouge">CreationUtcTime</code> is file creation metadata; use event time when reconstructing when this observation occurred. Further process context requires actual corresponding evidence, not a PID-only join.</p> <p><br /></p> <p><strong>Elastic Security: EQL Over Linux Defend Events</strong></p> <p>Use Elasticsearch EQL for these examples, not Kusto or the Kibana Query Language search bar. Select <code class="language-plaintext highlighter-rouge">logs-endpoint.events.process*</code> for ELA-A and <code class="language-plaintext highlighter-rouge">logs-endpoint.events.file*</code> for ELA-B, with Elastic Defend data from the selected Linux endpoints. Set a bounded seven-day or incident-specific <code class="language-plaintext highlighter-rouge">@timestamp</code> range outside the query text, and check result limits or truncation.</p> <p>The process selector follows the documented Defend shape in <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_suspicious_command_execution.toml">Elastic’s web-server command-execution rule</a>: a Linux process start with <code class="language-plaintext highlighter-rouge">event.action == "exec"</code>. In the <a href="https://www.elastic.co/docs/reference/ecs/ecs-process">ECS process fields</a>, <code class="language-plaintext highlighter-rouge">process.parent.*</code> describes the parent; <code class="language-plaintext highlighter-rouge">process.*</code> describes the child on this process event.</p> <pre><code class="language-eql">process where host.os.type == "linux" and host.name == "REPLACE_WITH_PHP_HOST" and event.type == "start" and event.action == "exec" and process.parent.name in ( "apache2", "httpd", "php-fpm", "php-fpm8.2", "php-fpm8.3", "php-fpm8.4", "php-fpm8.5", "php-cgi", "frankenphp", "lsphp" ) and process.name in ( "sh", "bash", "dash", "zsh", "ksh", "csh", "tcsh", "fish", "python", "python3", "perl", "ruby", "node", "php", "pwsh" ) </code></pre> <p>Keep the full matching event: timestamp, event ID, host identity, available user and container identity, process entity IDs, PID/start information, executable paths, command lines, and parent context. Don’t confuse a process name with the PHP script being handled. The query selects a relationship between recorded processes, not a route or request.</p> <p>For ELA-B, <code class="language-plaintext highlighter-rouge">process.*</code> on the qualified Defend file event is the responsible process. Elastic’s <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_php_file_creation_in_wordpress_plugin_dir.toml">PHP plugin-file rule supplies the creation/change pattern</a>; its <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_unusual_file_creation.toml">unusual web-server file rule also selects the rename action</a>. The latter is a <code class="language-plaintext highlighter-rouge">kuery</code>/<code class="language-plaintext highlighter-rouge">new_terms</code> rule. This sample is an EQL event filter, not a reproduction of that rule’s baseline behavior.</p> <pre><code class="language-eql">file where host.os.type == "linux" and host.name == "REPLACE_WITH_PHP_HOST" and (event.type in ("creation", "change") or event.action == "rename") and process.name in ( "apache2", "httpd", "php-fpm", "php-fpm8.2", "php-fpm8.3", "php-fpm8.4", "php-fpm8.5", "php-cgi", "frankenphp", "lsphp" ) and file.path like "/var/www/html/*" and file.name : ("*.php", "*.phtml", "*.php5", "*.phar") </code></pre> <p><a href="https://www.elastic.co/docs/reference/ecs/ecs-file">ECS separates the full <code class="language-plaintext highlighter-rouge">file.path</code> from <code class="language-plaintext highlighter-rouge">file.name</code></a>. The root uses case-sensitive <code class="language-plaintext highlighter-rouge">like</code>; the filename’s colon patterns are case-insensitive under <a href="https://www.elastic.co/docs/reference/query-languages/eql/eql-syntax">EQL’s documented operators</a>. Thus the intended suffix selection includes <code class="language-plaintext highlighter-rouge">.PHP</code>, but not <code class="language-plaintext highlighter-rouge">.php.jpg</code>, while a differently cased Linux root is different. These are candidate suffixes, not proof the server will execute the file. <a href="https://www.php.net/manual/en/install.unix.apache2.php">PHP handler mappings can associate other extensions with PHP</a>, so adapt them to the actual configuration.</p> <p>Don’t quietly extend this writer contract to every ECS integration. <a href="https://www.elastic.co/docs/reference/ecs/ecs-event"><code class="language-plaintext highlighter-rouge">event.action</code> is implementer-defined</a>, and the required mappings must exist. An unmapped query field can make EQL fail; hiding it with optional-field syntax doesn’t repair missing collection. Resolve that adoption problem before treating the query as a hunt you can interpret.</p> <p><br /></p> <p><strong>What a Match Actually Buys You</strong></p> <p><em>Editorial Note: The following update scenario and path are fictional, not an observed incident.</em></p> <p>Suppose an inventoried Linux <code class="language-plaintext highlighter-rouge">php-fpm8.3</code> worker launches <code class="language-plaintext highlighter-rouge">sh</code> for an approved rendering helper. During the same application update, the worker itself writes <code class="language-plaintext highlighter-rouge">/var/www/html/wp-content/plugins/report-helper/export.php</code>. Both observations fall within the selected hosts and time window. The process and file examples for MDE or Elastic can select them independently.</p> <p>Now add the context: the application owner authorized the update, the helper’s actual command matches its documented function, and the written file matches the approved plugin package. That supplies a coherent legitimate explanation for both observations. It isn’t merely a maintenance-window excuse. Elastic’s public guidance explicitly recognizes <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_suspicious_command_execution.toml">legitimate administration and conversion workflows</a> and <a href="https://raw.githubusercontent.com/elastic/detection-rules/main/rules/linux/persistence_webserver_php_file_creation_in_wordpress_plugin_dir.toml">plugin updates</a> as explanations to investigate.</p> <p>Two matching families don’t become independent proof of compromise when one authorized workflow explains both. Nor should that workflow become a blanket exclusion for every file written by the same service account. The relevant evidence is the particular command, content, authorization, and application function.</p> <p>Keep the endpoint observations unchanged, but remove the authorization record and the trusted package comparison. The correct result is unresolved provenance. Find out what the file contains, where it came from, whether the application should write it, and whether the helper invocation belongs to an authorized operation. A <code class="language-plaintext highlighter-rouge">.php</code> extension establishes none of those things. Even a file that turns out to be a backdoor doesn’t, on its own, identify the vulnerability that placed it there.</p> <p>For a child-process result, I’d start with the executable path, command line, parent instance, and application feature that could legitimately invoke it. For a file result, I’d preserve the content and original event, compare it with the expected deployment artifact, and inspect the recorded writer and rename history where available. <a href="https://www.microsoft.com/en-us/security/blog/2021/02/11/web-shell-attacks-continue-to-rise/">Microsoft’s process-tree guidance</a> is useful here because a filename alone is thin evidence.</p> <p>Then bring in whatever web and application records actually connect the activity to a route, account, or request. Process identity is not request identity. Timestamp proximity is a lead, not proof that a nearby access-log entry caused the process or file event. Where that connection isn’t recorded, leave it unresolved rather than filling it in from the worker’s name.</p> <p>Family B has a blind spot: if the worker starts <code class="language-plaintext highlighter-rouge">sh</code> and <code class="language-plaintext highlighter-rouge">sh</code> writes the file, the direct-worker writer filter won’t match that write. Family A may still find the shell child. Follow its actual recorded ancestry and file activity separately; adding every shell as a writer would discard the PHP provenance the filter was meant to preserve. Similarly, A doesn’t follow <code class="language-plaintext highlighter-rouge">worker -&gt; helper -&gt; interpreter</code> through an unlisted intermediate process.</p> <p><br /></p> <p><strong>Keep Patch Evidence and Hunt Evidence Separate</strong></p> <p>A no-match result is only interpretable with its scope attached: which hosts, which retained interval, which populated parent/writer fields, which executable names, and which roots. Missing collection, excluded events, different container paths, unlisted programs, existing files, or activity outside that interval can all leave these predicates with nothing to select. In-memory execution and PHP-only backdoor activity may produce neither selected behavior.</p> <p>The same restraint applies to error logs. The stale-state advisory says <a href="https://github.com/php/php-src/security/advisories/GHSA-6wmf-4rv5-p3rp">logs supplement a static configuration and application-flow audit</a>; absent errors aren’t proof of safety, including when historical logging excluded relevant diagnostics. Use logs to develop the investigation, not as a replacement for verifying the loaded fix.</p> <p>What I want from the response is a defensible answer to each original question. For remediation, show that the affected service is running the fixed PHP runtime, with the workers replaced or restarted. For investigation, show what behavior was found, what explains it, and what remains outside the evidence. Close the patch task on serving-runtime evidence. Don’t close the compromise question merely because a generic hunt returned no rows.</p> <p><img src="http://canarytokens.com/stuff/92y0da5x1x64inmcyqijsf69v/post.jsp" style="display: none;" /></p>
  115. Hunting VectraRAT With the Telemetry You Actually Have

    Mon, 19 Oct 2026 00:00:00 -0000

    I thought I’d throw out another analysis 101 post after reading up on VectraRAT. SOCRadar’s report ends with a useful hunting list. The report gives you a mutex, a config file, twelve hashes, and a list of C2 addresses. Let’s talk about how you hunt that, where each of those artifacts actually exists in your telemetry, all that good stuff.
    <p>I thought I’d throw out another analysis 101 post after reading up on VectraRAT. <a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar’s report</a> ends with a useful hunting list. The report gives you a mutex, a config file, twelve hashes, and a list of C2 addresses. Let’s talk about how you hunt that, where each of those artifacts actually exists in your telemetry, all that good stuff.</p> <p>Some orgs have documented fields. Some have only partial coverage. Others have no representation in the schemas you’re using.</p> <p>That distinction obviously matters. A malware report describes behavior; your data source determines which parts you can test. An empty query can mean no matching event, a missing collection prerequisite, or an artifact your telemetry cannot represent at all.</p> <p>So I thought I’d map SOCRadar’s reported VectraRAT artifacts to Microsoft Defender advanced hunting fields and Sysmon events, then working through three read-only hunts. The goal isn’t to claim coverage I haven’t measured. It’s to show what each query can establish, what it can’t, and how one might write up the difference.</p> <p><br /></p> <p><strong>What the report says</strong></p> <p><a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar’s Threat Research Unit published on September 14, 2026</a>, and <a href="https://www.darkreading.com/endpoint-security/vectrarat-hack-windows-enterprises">Dark Reading relayed the findings on the 15th</a>. That’s one malware-research lineage. Dark Reading adds outside commentary, but presents no independent examination of the malware. The behavioral findings here are SOCRadar’s. I did not run a sample or independently validate its behavior; the work here is public-source comparison and passive infrastructure checks.</p> <p>The reported architecture pairs a Go control server called VectraHub, with a Vue3 operator panel compiled in, with a native C++ Windows implant. They use a proprietary TCP protocol on port 3308 with a 5-byte header and MessagePack payloads. Rental starts at $250 a month.</p> <p>Delivery is somebody else’s problem. SOCRadar attributes the observed Amadey and ClickFix campaigns, including TurboTax lures, to buyers rather than the developer. It dates its first observation of live infrastructure to June 23, 2026, through an open directory. That is not a collection date for every published indicator: the investigation continued across other infrastructure and samples.</p> <p>The <a href="https://socradar.io/resources/whitepapers/vectrarat-threat-research-report/">69-page report requires a marketing form</a> but is worth a read.</p> <p><br /></p> <p><strong>The hash list has a coverage problem</strong></p> <p>The IOC section publishes 12 SHA-256 values. Before pasting them into a hunt, check the destination column.</p> <p>Microsoft documents <code class="language-plaintext highlighter-rouge">SHA256</code> and <code class="language-plaintext highlighter-rouge">InitiatingProcessSHA256</code> as usually unpopulated in <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-deviceprocessevents-table">DeviceProcessEvents</a> and <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicefileevents-table">DeviceFileEvents</a>, with advice to use SHA-1 when available. <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicenetworkevents-table">DeviceNetworkEvents</a> has <code class="language-plaintext highlighter-rouge">InitiatingProcessSHA256</code>, with the same warning; it does not have a file-level <code class="language-plaintext highlighter-rouge">SHA256</code> column.</p> <p>You can’t derive SHA-1 from SHA-256. Matching across those hash types needs a source that already maps both to the same file, such as an authorized intelligence feed or hash inventory. Nobody needs to fetch live malware to bridge that gap.</p> <p>Run the lookback over the available SHA-256 fields anyway, but scope it deliberately. A short window alone doesn’t guarantee a cheap query. “No match among events with the searched hash field populated” is defensible; “clean” isn’t. The file-table documentation also lists reasons a hash may be missing altogether, including remote storage, a lock held by another process, compression, or a file marked virtual.</p> <p>There’s another limit. <a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar describes a builder</a> that patches the C2 address into 256-byte null placeholder blocks and rewrites VERSIONINFO resources from Linux, recalculating the PE checksum. Changing configuration or metadata bytes can change the hash. Identical builds can still be distributed repeatedly, so the published values remain worth checking. They aren’t a family signature.</p> <p>The public list doesn’t map individual hashes to sample types or collection dates. Don’t assume all 12 are Windows implant files.</p> <p><br /></p> <p><strong>Which artifacts have a field, and which just have a paragraph</strong></p> <p>This compares <a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar’s reported artifacts</a> with the three schemas above and <a href="https://learn.microsoft.com/en-us/sysinternals/downloads/sysmon">Sysmon’s published event list</a>. The last column is what stops a plausible query becoming an implausible conclusion.</p> <p><img width="1400" height="860" alt="The key pillars of observability" src="https://github.com/user-attachments/assets/9e6d3f0e-c57c-43d9-8f7d-11d62b3a0ac7" /> <em>Credit to https://medium.com/%40imJenal/observability-deconstructed-a-comparative-study-of-logs-metrics-and-traces-d5bfe34feb8c</em></p> <table> <thead> <tr> <th>Reported artifact</th> <th>Documented field or event</th> <th>What a hit does not establish</th> </tr> </thead> <tbody> <tr> <td>Seven IP addresses, one delivery domain</td> <td><code class="language-plaintext highlighter-rouge">RemoteIP</code>, <code class="language-plaintext highlighter-rouge">RemoteUrl</code>, <code class="language-plaintext highlighter-rouge">RemotePort</code> in DeviceNetworkEvents; Sysmon event 3</td> <td>The reported roles include delivery and staging as well as C2. An address match doesn’t establish successful communication or infection</td> </tr> <tr> <td>Default PE metadata: Product and Company <code class="language-plaintext highlighter-rouge">Vectra</code>, description <code class="language-plaintext highlighter-rouge">Vectra client</code>, version <code class="language-plaintext highlighter-rouge">0.2</code></td> <td><code class="language-plaintext highlighter-rouge">ProcessVersionInfoCompanyName</code>, <code class="language-plaintext highlighter-rouge">ProcessVersionInfoProductName</code>, <code class="language-plaintext highlighter-rouge">ProcessVersionInfoFileDescription</code>, <code class="language-plaintext highlighter-rouge">ProcessVersionInfoProductVersion</code></td> <td>Version metadata isn’t signing evidence. SOCRadar’s builder description also makes these strings operator-editable</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">callback.json</code> in the temp directory</td> <td><code class="language-plaintext highlighter-rouge">FileName</code>, <code class="language-plaintext highlighter-rouge">FolderPath</code>, <code class="language-plaintext highlighter-rouge">ActionType</code> in DeviceFileEvents; Sysmon event 11, for create or overwrite</td> <td>A file event doesn’t reveal the JSON contents or establish that the implant read it</td> </tr> <tr> <td>Outbound TCP 3308</td> <td><code class="language-plaintext highlighter-rouge">RemotePort</code>, <code class="language-plaintext highlighter-rouge">Protocol</code>, initiating-process columns; Sysmon event 3</td> <td>These mapped fields don’t establish session duration or concurrency. Repeated events aren’t duration measurements</td> </tr> <tr> <td><code class="language-plaintext highlighter-rouge">winver.exe</code> launched and terminated, then <code class="language-plaintext highlighter-rouge">computerdefaults.exe</code> in the elevation path</td> <td>Process lineage, <code class="language-plaintext highlighter-rouge">ProcessIntegrityLevel</code>, <code class="language-plaintext highlighter-rouge">ProcessTokenElevation</code>; Sysmon events 1 for creation and 5 for termination</td> <td>This is an optional build branch. The reported final step declares a spoofed parent, so lineage needs corroboration</td> </tr> <tr> <td>The ntdll debug-object sequence</td> <td>No dedicated field in these three tables. Sysmon event 10 reports process access</td> <td>Opening another process isn’t comprehensive debug API tracing</td> </tr> <tr> <td>Mutex <code class="language-plaintext highlighter-rouge">LocalVectra.Client.SingleInstance</code></td> <td>No mutex field in these schemas or mutex event in Sysmon’s published list; events 17 and 18 cover named pipes</td> <td>A negative result through this mapping cannot test the mutex indicator</td> </tr> <tr> <td>Bursts of reads across <code class="language-plaintext highlighter-rouge">.env</code>, <code class="language-plaintext highlighter-rouge">.conf</code>, <code class="language-plaintext highlighter-rouge">.config</code></td> <td>DeviceFileEvents covers file creation, modification, and other file-system events</td> <td>The page doesn’t document general file-read coverage. File events don’t establish exfiltration</td> </tr> </tbody> </table> <p><em>Editorial Note: the source renders the strings as <code class="language-plaintext highlighter-rouge">LocalVectra.Client.SingleInstance</code> and <code class="language-plaintext highlighter-rouge">%TEMP%callback.json</code>. The separators are uncertain. I haven’t normalized either; the file query below matches only the basename.</em></p> <p>“No match in available fields” and “not represented by this mapping” are different results. If Sysmon is your source, the relevant event also needs to be enabled and collected. <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicenetworkevents-table">The Defender tables require Defender for Endpoint records</a>; a license alone doesn’t supply them.</p> <p>I haven’t measured column population in any environment. Actual coverage depends on collection and retained history, not the existence of a schema page. And this mapping says nothing about every other EDR’s capabilities.</p> <p><br /></p> <p><strong>Query one: keep the reported roles attached</strong></p> <p>Start with the address lookback, keeping SOCRadar’s role labels attached. An Amadey staging address and the reported primary C2 address shouldn’t land in the same bucket in the hunt notes. These are historical indicators to check against retention, not a verified current server map.</p> <p>All three queries below use published fields and have had a manual review. They haven’t been compiled or run against a Defender tenant. They’re read-only hunt starting points, not validated detections. Check the in-portal schema and your collection before interpreting an empty result.</p> <pre><code class="language-kusto">let LookbackWindow = 7d; let ReportedAddresses = datatable(DefangedAddress: string, ReportedRole: string) [ "86[.]109[.]75[.]168", "Primary C2 and panel, exposed directory", "86[.]109[.]75[.]161", "ClickFix distribution, resolves reported delivery domain", "178[.]16[.]54[.]148", "ClickFix panel serving VectraRAT and NetSupport", "195[.]20[.]115[.]77", "Secondary cluster, exposed panel and directory", "91[.]219[.]236[.]179", "Related infrastructure reached during pivoting", "91[.]92[.]242[.]236", "Amadey panel and VectraRAT staging", "195[.]63[.]145[.]106", "Shared TCP 3308 infrastructure with Vidar" ] | extend RemoteIP = replace_string(DefangedAddress, "[.]", "."); ReportedAddresses | join kind=inner ( DeviceNetworkEvents | where Timestamp &gt; ago(LookbackWindow) | where isnotempty(RemoteIP) | project Timestamp, DeviceId, DeviceName, NetworkAction = ActionType, RemoteIP, RemotePort, Protocol, InitiatingProcessFileName, InitiatingProcessFolderPath, InitiatingProcessUniqueId ) on RemoteIP | project Timestamp, DeviceName, DeviceId, RemoteIP, ReportedRole, NetworkAction, RemotePort, Protocol, InitiatingProcessFileName, InitiatingProcessFolderPath, InitiatingProcessUniqueId | top 200 by Timestamp asc </code></pre> <p><code class="language-plaintext highlighter-rouge">replace_string</code> refangs the static inputs inside the query; it doesn’t contact the addresses. Choose an authorized lookback your retention supports. Seven days is an example, and certainly not a promise of low cost or coverage back to June.</p> <p><code class="language-plaintext highlighter-rouge">NetworkAction</code> preserves <code class="language-plaintext highlighter-rouge">ActionType</code> in all three examples. <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-devicenetworkevents-table">DeviceNetworkEvents covers connections and related events</a>, so check your tenant’s documented action values for success and direction. A <code class="language-plaintext highlighter-rouge">RemoteIP</code> hit is an event associated with a published address, not proof of communication or infection.</p> <p><br /></p> <p><strong>Query two: default metadata plus network activity from the same process</strong></p> <p>Version <code class="language-plaintext highlighter-rouge">0.2</code> on its own is weak. Pairing the reported name strings with that version makes a more useful lead, especially when the same process has network events. That means joining on identity rather than hoping a PID holds still.</p> <p><a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-deviceprocessevents-table">Microsoft defines <code class="language-plaintext highlighter-rouge">ProcessUniqueId</code> as the Windows Process Start Key</a>; the network table exposes the initiating process’s key as <code class="language-plaintext highlighter-rouge">InitiatingProcessUniqueId</code>. Aliasing the former to the latter and joining with <code class="language-plaintext highlighter-rouge">DeviceId</code> avoids a PID-only match. Sysmon’s <code class="language-plaintext highlighter-rouge">ProcessGUID</code> serves a similar correlation purpose when Windows reuses PIDs.</p> <pre><code class="language-kusto">let LookbackWindow = 7d; let DefaultMetadataProcesses = DeviceProcessEvents | where Timestamp &gt; ago(LookbackWindow) | where ProcessVersionInfoCompanyName =~ "Vectra" or ProcessVersionInfoProductName =~ "Vectra" or ProcessVersionInfoFileDescription =~ "Vectra client" | where ProcessVersionInfoProductVersion == "0.2" or ProcessVersionInfoProductVersion startswith "0.2." | where isnotempty(DeviceId) and isnotempty(ProcessUniqueId) | project ProcessTimestamp = Timestamp, DeviceId, DeviceName, AccountName, FileName, FolderPath, ProcessCommandLine, ProcessIntegrityLevel, ProcessTokenElevation, ProcessVersionInfoCompanyName, ProcessVersionInfoProductName, ProcessVersionInfoProductVersion, InitiatingProcessUniqueId = ProcessUniqueId; DefaultMetadataProcesses | join kind=inner ( DeviceNetworkEvents | where Timestamp &gt; ago(LookbackWindow) | where isnotempty(DeviceId) and isnotempty(InitiatingProcessUniqueId) | project NetworkTimestamp = Timestamp, DeviceId, NetworkAction = ActionType, RemoteIP, RemotePort, Protocol, InitiatingProcessUniqueId ) on DeviceId, InitiatingProcessUniqueId | where NetworkTimestamp &gt;= ProcessTimestamp | project ProcessTimestamp, NetworkTimestamp, DeviceId, DeviceName, AccountName, FileName, FolderPath, ProcessIntegrityLevel, ProcessTokenElevation, ProcessVersionInfoCompanyName, ProcessVersionInfoProductName, ProcessVersionInfoProductVersion, NetworkAction, RemoteIP, RemotePort, Protocol, InitiatingProcessUniqueId | top 200 by NetworkTimestamp asc </code></pre> <p>The name filter accepts a match in any one of the company, product or description fields. The separate version filter must also match. It accepts exactly <code class="language-plaintext highlighter-rouge">0.2</code> or a <code class="language-plaintext highlighter-rouge">0.2.</code> prefix, a formatting assumption that accommodates a value such as <code class="language-plaintext highlighter-rouge">0.2.0.0</code> without also taking <code class="language-plaintext highlighter-rouge">0.20</code>. I haven’t observed those variants in data.</p> <p>The chronology filter requires a network event at or after the retained process event’s <code class="language-plaintext highlighter-rouge">Timestamp</code>. There is no process <code class="language-plaintext highlighter-rouge">ActionType</code> filter here, so that timestamp isn’t necessarily the process’s first creation time. The query also requires a matching process row inside the lookback. An older-running process with no such row will be missed.</p> <p>This query and the next only pair events attributed to the matching process itself. If a child makes the connection, following that needs another hop through process lineage.</p> <p>Both joins discard empty device or process identities on each side. Count those exclusions when assessing the hunt’s inputs, but don’t turn excluded-row counts into a percentage of devices covered. That’s a different denominator.</p> <p><br /></p> <p><strong>Query three: a file event, then network activity</strong></p> <p>The C2 override is the artifact I find most interesting operationally. <a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar describes the implant’s startup check</a> for <code class="language-plaintext highlighter-rouge">callback.json</code> in <code class="language-plaintext highlighter-rouge">%TEMP%</code>: if it exists and contains <code class="language-plaintext highlighter-rouge">"callback": true</code>, the implant takes its C2 address from the file instead of the PE’s embedded configuration. In the reported panel-driven flow, an operator requests the change and the override is written on the endpoint. I haven’t reproduced that flow.</p> <p><img width="1448" height="1086" alt="Same-Process Correlation Explained" src="https://github.com/user-attachments/assets/c42d348b-7d11-47c7-8dda-cf9c5749b6a8" /></p> <p>That isn’t the only possible origin of a file with this name. <code class="language-plaintext highlighter-rouge">callback.json</code> is generic, and an infected host where the override wasn’t used may not have it at all. The query looks for a file event followed by network activity attributed to the same process, without assuming the file was created or its contents used. This is a same-process correlation, and not a claim that the file controls the connection. Specifically, it prioritizes a sequence for investigation. It does <em>not</em> establish that the process read the file or used its contents.</p> <pre><code class="language-kusto">let LookbackWindow = 7d; let CorrelationWindow = 10m; let CallbackFileEvents = DeviceFileEvents | where Timestamp &gt; ago(LookbackWindow) | where FileName =~ "callback.json" | where isnotempty(DeviceId) and isnotempty(InitiatingProcessUniqueId) | project FileTimestamp = Timestamp, DeviceId, DeviceName, FileAction = ActionType, FolderPath, InitiatingProcessFileName, InitiatingProcessFolderPath, InitiatingProcessCommandLine, InitiatingProcessUniqueId; CallbackFileEvents | join kind=inner ( DeviceNetworkEvents | where Timestamp &gt; ago(LookbackWindow) | where isnotempty(DeviceId) and isnotempty(InitiatingProcessUniqueId) | project NetworkTimestamp = Timestamp, DeviceId, NetworkAction = ActionType, RemoteIP, RemotePort, Protocol, InitiatingProcessUniqueId ) on DeviceId, InitiatingProcessUniqueId | where NetworkTimestamp &gt;= FileTimestamp | where NetworkTimestamp &lt;= FileTimestamp + CorrelationWindow | project FileTimestamp, NetworkTimestamp, DeviceId, DeviceName, FileAction, FolderPath, InitiatingProcessFileName, InitiatingProcessFolderPath, NetworkAction, RemoteIP, RemotePort, Protocol, InitiatingProcessUniqueId | top 200 by FileTimestamp asc </code></pre> <p>Ten minutes is an analyst’s starting window, not a measured property of VectraRAT. Delayed activity from the same process can fall outside it; widen the interval if the investigation calls for it. Widening cannot bridge different process identities across a reboot.</p> <p><code class="language-plaintext highlighter-rouge">kind=inner</code> keeps matching pairs rather than deduplicating left-side keys as the default <code class="language-plaintext highlighter-rouge">innerunique</code> join would. One process can therefore produce multiple file/network pairs. The retained <code class="language-plaintext highlighter-rouge">FileAction</code> and <code class="language-plaintext highlighter-rouge">NetworkAction</code> values let you inspect the event types instead of assuming creation followed by a successful connection.</p> <p>Nor does the sequence establish a destination change. That needs retained network history for comparison. The file event supplies no JSON contents, so it can’t establish configuration use either.</p> <p>The <code class="language-plaintext highlighter-rouge">top 200</code> in each example returns at most 200 rows, earliest by its sort column. It limits the displayed result, not the work required by the join or the total matching events. For operational use, narrow the time slices and retrieve all matches before drawing coverage conclusions; don’t mistake the cap for a complete handover.</p> <p><br /></p> <p><strong>The port is context, not a verdict</strong></p> <p>TCP 3308 can help prioritize results, but where you apply it matters. SOCRadar reports it as the C2 port; the delivery and staging addresses weren’t all claimed to serve C2 there. Filtering the indicator query on 3308 would drop other reported roles.</p> <p>And a port isn’t a family signature. It’s a configuration choice. These mapped fields contain no protocol payload, so the reported 5-byte header and MessagePack framing are outside this layer.</p> <p>If <a href="https://learn.microsoft.com/en-us/sysinternals/downloads/sysmon">Sysmon supplies your network events</a>, event 3 needs to be enabled and configured. Microsoft’s default-install example uses SHA-1 image hashing and no network monitoring. Those are Sysmon’s defaults, not a statement about Defender’s separate data path. Collection volume depends on the filtering you choose; I haven’t measured it. Know what your selected source records before writing “no evidence of C2 activity.”</p> <p><br /></p> <p><strong>What I could not confirm about the infrastructure</strong></p> <p>I tried to turn the embedded Vue3 panel into a reusable search fingerprint. The passive checks didn’t validate one. They didn’t prove that one cannot exist.</p> <p>On September 18, 2026, InternetDB lookups returned records for all seven published addresses, with none listing TCP 3308. <a href="https://internetdb.shodan.io/openapi.json">InternetDB’s response schema</a> provides no observation timestamp or port history. This describes the returned records, not host liveness or a current C2 map; some addresses have delivery or related roles anyway. No reported host was contacted.</p> <p>The <a href="_pipeline/research_out/vectrarat_passive_checks_2026-09-18.json">saved urlscan.io searches</a> found zero VectraHub title matches and zero primary-page IP matches for three reported panel or related seeds, with date bounds of June 1 to September 19, 2026. A separate Nyxel title search returned 12 name matches, none corroborated as VectraRAT infrastructure. The <a href="https://urlscan.io/docs/search/">title and primary-page fields</a> don’t provide complete coverage of resource contacts or panel exposure.</p> <p>The automated Shodan VectraHub search showed no results, while the filtered storage-key attempt stopped at a login requirement. That’s an access restriction, not a zero-match result. I also got no hits in the manual Shodan and Censys follow-up. <a href="https://help.shodan.io/the-basics/search-query-fundamentals">Shodan’s default search covers collected banner <code class="language-plaintext highlighter-rouge">data</code></a>, so it doesn’t necessarily include a separately served JavaScript bundle.</p> <p>The two Nyxel storage keys in SOCRadar’s report remain a possible, untested route. <a href="https://urlscan.io/docs/search/">urlscan’s <code class="language-plaintext highlighter-rouge">content.storageNames</code> field</a> requires Professional-tier access or higher and indexes storage items set by a page. A string in JavaScript doesn’t prove an unauthenticated page sets that item.</p> <p>No verified TLS certificate, favicon hash or distinctive HTTP asset hash emerged as a seed. The raw protocol remains untested, including whether TLS is used. There is no known certificate here to support a Certificate Transparency pivot. A corroborated panel artifact would be needed before calling this a reusable fingerprint. SOCRadar’s ownership inference from a shared /24 and provider remains an inference; these checks don’t confirm it.</p> <p><em>Editorial Note: <a href="_pipeline/research_out/vectrarat_passive_checks_2026-09-18.json">The September 18 checkpoint</a> records the automated requests and outcomes. The manual Shodan and Censys results are author-reported, not independently reproduced; their exact queries weren’t captured. All collection described here was passive retrieval of existing third-party records or public documentation.</em></p> <p><br /></p> <p><strong>Artifacts</strong></p> <ul> <li>IP addresses and reported roles: the defanged datatable in query one</li> <li>Delivery domain: <code class="language-plaintext highlighter-rouge">verify-cloud[.]digital</code></li> <li>Mutex, as rendered in the source: <code class="language-plaintext highlighter-rouge">LocalVectra.Client.SingleInstance</code></li> <li>C2 override file, as rendered in the source: <code class="language-plaintext highlighter-rouge">%TEMP%callback.json</code></li> <li>Default stub metadata: Product <code class="language-plaintext highlighter-rouge">Vectra</code>, Company <code class="language-plaintext highlighter-rouge">Vectra</code>, description <code class="language-plaintext highlighter-rouge">Vectra client</code>, version <code class="language-plaintext highlighter-rouge">0.2</code></li> <li>C2 port: TCP 3308. Panel ports: TCP 8080, 8888. Auxiliary: TCP 4457, 6667</li> <li>The 12 SHA-256 values remain in <a href="https://socradar.io/blog/vectrarat-undocumented-stack-maas/">SOCRadar’s IOC section</a>, intact and copy-pasteable</li> </ul> <p><br /></p> <p><strong>Working a hit, and writing up a miss</strong></p> <p>Start a query-one hit with <code class="language-plaintext highlighter-rouge">NetworkAction</code> and the address’s reported role. Don’t treat a failed attempt as proof of communication. On the returned <code class="language-plaintext highlighter-rouge">DeviceId</code>, review the initiating image and its execution context before treating the address association as family evidence.</p> <p>For query two, use file evidence already available through your authorized investigation. Editable version strings don’t answer the signing question. <a href="https://learn.microsoft.com/en-us/defender-xdr/advanced-hunting-deviceprocessevents-table">The DeviceProcessEvents page’s example joins the created process’s <code class="language-plaintext highlighter-rouge">SHA1</code> to certificate information</a>, rather than using the initiating process’s signature status as a substitute.</p> <p>For query three, inspect <code class="language-plaintext highlighter-rouge">FileAction</code> and <code class="language-plaintext highlighter-rouge">FolderPath</code>, then any file contents available through that investigation. Compare the process’s destinations with retained history. The sequence isn’t automatic family confirmation; escalate according to the actual evidence.</p> <p>If the hunts return nothing, record the queried dates and which fields were populated. Separate a zero-match result from missing collection prerequisites. Different builds can evade these published hashes, and the joins still need usable identity fields and a matching record inside the window. None of that clears the environment.</p> <p>Leave artifacts not represented by this published mapping as named follow-up work. Missing telemetry is a troubleshooting task, not a reason to dismiss evidence that you do have.</p> <p><img src="http://canarytokens.com/about/epl1cn7v5e12j5sxbsge1sopo/index.html" style="display: none;" /></p>
  116. Srsly Risky Biz: Russia's Digital Detox Boomerang

    Thu, 15 Oct 2026 04:11:16 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://prowler.com/"><em><u>Prowler</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/2026-10-15-Yandex_main_office.jpg" alt="Srsly Risky Biz: Russia&apos;s Digital Detox Boomerang"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://prowler.com/"><em><u>Prowler</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB186.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB186/&quot;&gt;Srsly Risky Biz: Russia&apos;s digital detox boomerang&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 20:29 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Russia&apos;s digital detox boomerang&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-5c7eff54-265a-43fb-a3e8-bf6a455fc8a6.jpeg" class="kg-image" alt="Srsly Risky Biz: Russia&apos;s Digital Detox Boomerang" loading="lazy" width="2000" height="1500" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/10/data-src-image-5c7eff54-265a-43fb-a3e8-bf6a455fc8a6.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/10/data-src-image-5c7eff54-265a-43fb-a3e8-bf6a455fc8a6.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/10/data-src-image-5c7eff54-265a-43fb-a3e8-bf6a455fc8a6.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/10/data-src-image-5c7eff54-265a-43fb-a3e8-bf6a455fc8a6.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Yandex Main Office, WikiFido on Wikimedia Commons</span></figcaption></figure><p>Ukrainian drone strikes against Russian data centres are causing widespread disruption to the country&apos;s internet services.&#xa0;</p><p>Ukrainian President Volodymyr Zelensky <a href="https://x.com/ZelenskyyUa/status/2108509610990579862"><u>said the strikes</u></a>, targeting the data centres of Russian technology company Yandex, were a &quot;symmetrical response&quot; to <a href="https://therecord.media/kyiv-internet-providers-report-outages-after-russian-strikes"><u>Russian attacks</u></a> on several Ukrainian data centres that began in mid September. The Russian Ministry of Defense <a href="https://meduza.io/en/news/2026/10/08/russia-called-its-strikes-on-ukrainian-data-centers-a-forced-digital-detox-zelensky-says-kyiv-has-answered-by-hitting-a-data-center-belonging-to-russian-tech-giant-yandex"><u>called its strikes</u></a> on Ukrainian infrastructure a &quot;digital detox&quot;.&#xa0;</p><p>These tit-for-tat kinetic attacks have been far more impactful than either country&apos;s cyber operations. But Russia is more vulnerable to this type of disruption than Ukraine because of its digital sovereignty strategy.</p><p>At the time of writing, <a href="https://www.rferl.org/a/zelenskyy-says-yandex-strikes-crippled-pro-kremlin-bots-but-tech-experts-call-evidence-weak/33875054.html"><u>Yandex&apos;s three largest data centres</u></a> (the company operates six in total) have been struck. The first was hit on Wednesday last week. <a href="https://x.com/ZelenskyyUa/status/2109602415397749044"><u>Speaking to the Warsaw Security Forum</u></a>, President Zelensky claimed the strikes had led to a decline in Russian disinformation and propaganda bot activity. He said that &quot;it suddenly became a little easier to breathe on many social media platforms.&quot;</p><p>&quot;There was less hate, less division, and fewer provocations because there were fewer bots. And it&apos;s pretty clear where those bots were coming from.&quot;</p><p>There has been anecdotal <a href="https://x.com/unusual_whales/status/2109684604504555830"><u>support</u></a> <a href="https://x.com/MalcontentmentT/status/2109386031745364111"><u>online</u></a> for <a href="https://x.com/SpencerHakimian/status/2109948428075594097"><u>this narrative</u></a>, but unfortunately, <a href="https://x.com/vatnik_soup/status/2109542508682248478"><u>the objective data</u></a> we <a href="https://x.com/valdemarosted/status/2109742982568313237?s=46&amp;t=qPQ5BqXGA7aOu3w2fLYtww"><u>have found so far</u></a> indicates that Russian propaganda bots are <a href="https://x.com/antibot4navalny/status/2110035660715466895"><u>still out in full force</u></a>.&#xa0;</p><p>The quality of online discourse aside, Ukraine&apos;s strikes are having real impacts within Russia, and on Monday <em>The</em> <em>Moscow Times</em> reported that they <a href="https://www.themoscowtimes.com/2026/10/12/ukraines-strikes-on-yandex-data-centers-explained-a93953"><u>were causing</u></a> &quot;major disruptions in daily life&quot;.</p><p>Yandex is basically Russia&apos;s Google, AWS and Uber rolled into one. It provides payment and online map services, email, and a taxi booking and food delivery app. The disruptions went beyond its own internal services, too. Businesses running on Yandex Cloud, including grocery, fast-food and retail organisations were also affected.</p><p>With the Yandex Pay service disrupted, many customers were not able to check balances, make payments or transfer money. Dima Pirogov, the CEO of Yandex&apos;s FinTech division attempted to reassure customers <a href="https://t.me/yandex/5903"><u>via a statement on social media</u></a>.&#xa0;</p><p>&quot;All your money is safe. The money itself isn&apos;t stored in the data center.&quot;&#xa0;</p><p>Pirogov did not provide an expected restoration time, but said &quot;we won&apos;t stop until the service is restored&quot;. So, you can&apos;t have your money, but it is safe. Trust me bro&#x2026;</p><p>It&apos;s fair to say Yandex isn&apos;t having a great time this week, but Russia&#x2019;s attacks on Ukrainian infrastructure and data centres, which kicked this whole thing off, were also impactful. <a href="https://www.reuters.com/business/aerospace-defense/russia-hits-ukraines-largest-mobile-provider-strikes-data-centres-2026-09-27/"><u>Attacks on</u></a> telecommunications and internet service providers caused outages to around 100,000 households, <a href="https://t.me/mintsyfra/8476"><u>according to</u></a> Ukraine&apos;s digital transformation ministry. These attacks <a href="https://www.datacenterdynamics.com/en/news/russia-claims-strikes-on-de-novo-data-center-in-kyiv-ukraine/"><u>included a strike</u></a> on one of Ukraine&apos;s largest data centres.&#xa0;&#xa0;</p><p>Since mid-September this year the Russian military has <a href="https://dev.ua/en/news/cherez-masovani-ataky-rf-u-kyivstar-postrazhdaly-dva-data-tsentry-1790864621"><u>launched a series</u></a> of conventional attacks against Kyivstar targeting its data centres, a warehouse and <a href="https://www.datacenterdynamics.com/en/news/russian-drones-hit-kyivstar-hq-vodafone-data-center-in-ukraine/"><u>central offices</u></a>. These locations are all points of vulnerability that could plausibly result in significant impacts on Kyivstar.</p><p>Physically destroying mobile networks is a big deal, but Ukraine has some mitigations. Kyivstar users <a href="https://www.lightreading.com/satellite/lessons-from-a-war-zone-how-ukraine-s-kyivstar-harnesses-starlink-to-keep-users-online"><u>can already use</u></a> direct-to-device (D2D) satellite services for texting via Starlink, for example. And many critical Ukrainian government services are already hosted outside of the country including <a href="https://news.microsoft.com/source/emea/features/microsoft-extends-technology-support-ukraine-2027/"><u>in Microsoft</u></a> and <a href="https://aws.amazon.com/blogs/publicsector/ukraines-digital-transformation/"><u>Amazon&apos;s cloud</u></a> infrastructure. That shift to cloud services occurred rapidly after the start of the war.&#xa0;</p><p>The Ukrainians have &quot;lift and shift&quot; down to a fine art. One of Ukraine&apos;s largest data centres, the 4 MW capacity <a href="https://dmc-education.com/2026/09/27/ukraine-data-centres-2026-top-data-centres-capacity-mw-establishment-date-locations/"><u>Parkovyi facility in Kyiv</u></a>, was destroyed by Russian strikes in late September this year. By October 1 the <a href="https://www.facebook.com/ParkovyiDC/posts/pfbid02uPkULxGyjs8mQ1nXQzTUtMgAWRfWnWR7QZ3ojRaneHy1U35kSjMUnKrw2EaqhQoUl?__cft__[0]=AZi3UDl-ktFLEUdEwUQv5DVvXGdru3Ho4BcGBEJC93rO2R8yloakT3YC1b3bWkUXW5RcPf-2qkKzbsvDTXpa1Jy948TlaESFtwT78kFwOIKyeyMINKsg84leaUybyvTEbKF2tr163MRGrWrjAaU3WgQC0V3El3Wr&amp;__tn__=%2CO%2CP-R"><u>company said</u></a> that customer data was already in backup sites within the EU.</p><p>Russia&apos;s digital sovereignty efforts, on the other hand, puts the country at a structural disadvantage.&#xa0;</p><p>These efforts <a href="https://nestcentre.org/digital-sovereignty-or-digital-dictatorship/"><u>have been motivated</u></a> by the desire to maintain political control. Over the last two decades Russia has <a href="https://neweasterneurope.eu/2025/12/29/russias-next-step-in-the-sovereign-internet-towards-full-isolation/"><u>rolled out a series of measures</u></a> that have increased the government&apos;s capability for domestic censorship and surveillance.&#xa0;</p><p><a href="https://www.atlanticcouncil.org/in-depth-research-reports/issue-brief/reassessing-runet-russian-internet-isolation-and-implications-for-russian-cyber-behavior/"><u>One explicit goal</u></a> of this sovereignty push has been to develop the capability to disconnect Russia from the global internet. This has resulted in strict data localisation requirements. Russia has deliberately put all its eggs into the domestic data centre basket.&#xa0;</p><p>So it&apos;s bad news when these data centres get bombed. Russian media outlet <em>Kommersant</em> <a href="https://www.kommersant.ru/doc/9024163"><u>reports that</u></a> Yandex, is transferring some services to other domestic providers such as VK Cloud and Selectel.</p><p>That is just shifting eggs around the same basket though. Nikolay Mitrokhin, a researcher at Bremen University&apos;s Center for East European Studies said, <a href="https://www.rferl.org/a/ukraine-strikes-data-centers-disruption-shopping-business-russia/33875869.html"><u>as reported by</u></a> <em>Radio Free Europe</em>, that 95 percent of Russia&apos;s data centres are within range of Ukrainian drones.</p><p>Kyiv, he said, had &quot;every capability to drive Russia back into the analogue era&quot;.&#xa0;&#xa0;</p><p>By targeting data centres with real-world attacks, Russia opened itself up to retaliatory strikes that have done serious damage to the country&apos;s localised digital infrastructure. Data centres in both countries are equally vulnerable to bombs. But it&apos;s much harder for Russia to mitigate the impacts.&#xa0;</p><p>It probably should have stuck to cyber mayhem.</p><h2 id="why-it%E2%80%99s-time-to-ditch-ai-cyber-safeguards">Why It&#x2019;s Time to Ditch AI Cyber Safeguards</h2><p>Removing safeguards from open-weight AI models is easy, according to <a href="https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities"><u>a new report</u></a> from Anthropic, thus confirming what <a href="https://news.risky.biz/srsly-risky-biz-open-weight-model-advances-make-the-mythos-debate-moot/"><u>we&apos;ve said before</u></a>: advanced cyber capabilities will soon be available to all and sundry.</p><p>The report, released in late September, compares Chinese company Z.ai&apos;s newest model, GLM-5.3, with Anthropic&apos;s own models from both a cyber benchmark and safety perspective.</p><p>In short, GLM-5.3 is the most capable open-weight model from a cyber perspective, and Anthropic says it is relatively easy to overcome its safeguards &quot;with a variety of simple techniques&quot;.</p><p>One technique Anthropic examined is known as &quot;<a href="https://arxiv.org/abs/2406.11717"><u>abliteration</u></a>&quot;, which alters a model so that its built-in safeguards are removed. It took Anthropic 2,200 GPU hours of computation, at a cost of about USD&#x24;4,400, to abliterate GLM-5.3. That&apos;s not nothing, but Anthropic says several developers publicly released abliterated versions of GLM-5.3 within days of its release. So, anyone who wants to play with an unrestricted version doesn&apos;t even need to spend the money.</p><p>It&apos;s important to mention here that abliteration doesn&apos;t substantially affect a model&apos;s benchmark performance. Abliterated models do just fine when put through their paces, even if they lose a couple of points here and there.</p><p>Easier and cheaper safeguard bypass techniques also worked pretty well, according to Anthropic&apos;s research. Simply using a deceptive prompt, such as telling the model it was a red-teamer, worked 64 percent of the time. Gaslighting a model into thinking it had already considered, and decided to proceed, with a user&#x2019;s request had a 92 percent success rate.</p><p>Unsurprisingly, Anthropic found that Claude&apos;s safeguards worked 100 percent of the time. (We ran a similar project. We evaluated all the cybersecurity newsletters and determined that <em>Seriously Risky Business</em> is the best, 100 percent of the time!)</p><p>The messaging in the report isn&apos;t exactly subtle. Claude, good. Chinese open-weight models, bad.</p><p>This, together with Anthropic CEO Dario Amodei&apos;s <a href="https://darioamodei.com/post/we-must-pace-the-frontier"><u>essay</u></a> about pacing the frontier makes it pretty clear to us that the company wants some sort of international AI regulatory regime that China buys into.</p><p>In late September, Trump and Chinese President Xi Jinping did <a href="https://www.whitehouse.gov/fact-sheets/2026/09/fact-sheet-president-donald-j-trump-advances-a-fair-and-reciprocal-relationship-with-china-while-hosting-historic-state-visit/"><u>agree to</u></a> set up a bilateral channel for AI incidents. We think that&apos;s about as far as international cooperation on AI regulation is going to go for now.&#xa0;</p><p>As <a href="https://www.abc.net.au/news/2026-09-15/trump-says-a-high-iq-president-is-the-only-guardrail-ai-needs/107153326"><u>recently as late September</u></a> President Donald Trump appeared to categorically rule out stricter domestic regulation. Until he&apos;s on board, it&apos;s a non-starter.</p><p>So, where does this leave us?</p><p>We think it&apos;s only a matter of time before open-weight models provide the world with cyber capability that tightly-leashed American models don&apos;t. Everyone having access to advanced cyber capabilities definitely presents risks, and Bad Cyber Things&#x2122; will happen.&#xa0;</p><p>The practical policy response here is to loosen safeguards on American frontier models. It&#x2019;s time to make the most advanced capabilities broadly available to attackers and defenders, and let the chips land where they may.&#xa0;</p><p>It&apos;s time to let the AI cage fight begin.</p><p><em>Watch Amberleigh Jack and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/E7J8LDRABAk?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Russia&apos;s digital detox boomerang"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>North Korean laptop farm action:</strong> On Tuesday the US Department of Justice <a href="https://www.justice.gov/opa/pr/justice-department-announces-coordinated-actions-combat-north-korean-remote-it-worker"><u>announced</u></a> that it had carried out &quot;coordinated actions&quot; to disrupt fraudulent North Korean IT worker schemes. The actions included two arrests, searches of more than 40 suspected laptop farms across Texas, and seizing more than 190 devices.&#xa0;</li><li><strong>AI to help open source, critical infrastructure: </strong>Last week <a href="https://www.anthropic.com/news/anthropic-cyber-mission"><u>Anthropic launched</u></a> the Anthropic Cyber Mission which it described as a &quot;long-term commitment to secure the systems everyone depends on&quot;. We like the idea of various AI companies fighting each other to generate the most good publicity by making public infrastructure more secure.&#xa0;</li><li><strong>Oxygen Forensics dropped like a hot potato</strong>: <em>Politico </em><a href="https://www.politico.eu/article/european-police-agencies-drop-russian-software-company/"><u>reports</u></a> that multiple European law enforcement agencies will stop using Oxygen Forensics, a digital forensics company whose software was used to extract and analyse data from mobile phones in criminal investigations. This is fallout from the US Department of Justice <a href="https://www.justice.gov/usao-cdca/pr/tech-ceo-russian-national-arrested-complaint-alleging-they-hid-russian-ownership-and"><u>alleging in September</u></a> that the firm&apos;s CEO claimed the company was American and that its software was developed in the US, whereas it was actually a Russian firm.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Prowler&#x2019;s founder and CEO Toni de la Fuente about how AI now allows pretty much anyone to vibe-code &#x201c;killer apps&#x201d;. These custom apps are often internal for organisations and hosted on the cloud.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI146.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI146/&quot;&gt;Sponsored: AI is attacking&#x2026; it needs to defend, too&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 18:31 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: AI is attacking&#x2026; it needs to defend, too&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> examine how ShinyHunters and the FBI both deceived themselves into believing they were immune to cyber threats.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN185.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN185/&quot;&gt;Between Two Nerds: The FBI and ShinyHunters are the same&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 35:22 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: The FBI and ShinyHunters are the same&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/oKSnzUKPNAE?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Why the FBI and Shiny Hunters are the same"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Spyware chip found in Ledger crypto-wallets:</strong> Ledger, a maker of hardware-based cryptocurrency wallets, has <a href="https://x.com/Ledger_Support/status/2108551100613714002"><u>told customers</u></a> to discard and replace devices bought from a Southeast Asian reseller named CryptoBilis.</p><p>Customers were also told to move funds stored on those wallets to new blockchain addresses as soon as possible, to avoid having the funds stolen.</p><p>The company&apos;s warning came after blockchain security firm Tibane Labs published a <a href="https://www.tibane.net/research/ledger-nano-x-implant"><u>report</u></a> over the weekend detailing spy chips found inside Ledger Nano X wallets over the past year.</p><p>The spy chips were found soldered between the cryptographically secure element that handles cryptocurrency operations and the device&apos;s screen.</p><p>[<a href="https://news.risky.biz/risky-bulletin-spyware-chip-found-in-ledger-crypto-wallets/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>ShinyHunters suspects arrested in US and Jordan: </strong>The FBI has arrested a suspected associate of the ShinyHunters hacking group in Pennsylvania. The individual has been identified as Edward Dubrovsky, a Canadian national and cybersecurity expert. Dubrovsky was arrested in Pennsylvania last week. FBI Director Kash Patel claims he was connected to the recent hack of the FBI&apos;s job portal last month. [<a href="https://www.nytimes.com/2026/10/09/us/politics/fbi-hack-shinyhunters-arrest.html"><em><u>NYT</u></em></a> // <a href="https://x.com/FBIDirectorKash/status/2108566367011999780"><u>Kash Patel</u></a> // <a href="https://www.linkedin.com/in/edubrovs/"><u>Ed Dubrovsky on LinkedIn</u></a>]</p><p>In Jordan authorities have detained another suspected member of the ShinyHunters hacking group. Saif &#x200c;al-Din Khader is allegedly a ShinyHunters member known online as &quot;Rey.&quot; According to Reuters, Khader is allegedly cooperating with the FBI to track other members. [<a href="https://www.reuters.com/world/middle-east/key-shinyhunters-hacker-detained-jordan-is-cooperating-sources-say-2026-10-03/"><em><u>Reuters</u></em></a>]</p><p><strong>Authorities dismantle KillSec group, arrest members across Europe:&#xa0;</strong></p><p>European law enforcement agencies have cracked down and dismantled the KillSec ransomware group in a coordinated operation that seized servers, conducted several house searches across four countries, and detained three of the group&apos;s members.</p><p>The biggest arrest was a 16-year-old Romanian national living in Alicante, Spain, identified as the group&apos;s main administrator. The teen&apos;s name was not released, according to Spanish child privacy laws.</p><p>Two other suspects were arrested in Romania and the UK. The suspect arrested in the UK was identified as Fouad Eltibrizi, a Dutch national. The US has filed an extradition request for Eltibrizi, seeking him for a cyberattack against a Puerto Rico company in March 2025.</p><p>[<a href="https://news.risky.biz/risky-bulletin-authorities-dismantle-killsec-group-arrest-members-across-europe/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  117. When Is an Automated SOC Response Ready to Run?

    Mon, 12 Oct 2026 00:00:00 -0000

    Suppose you have a detection you believe. Several independent signals agree, the analysis holds up, and you’re no longer arguing with yourself about whether the host is compromised. Your tooling, or an agent operating on top of it, proposes to isolate the host from the network. You hesitate anyway.
    <p>Suppose you have a detection you believe. Several independent signals agree, the analysis holds up, and you’re no longer arguing with yourself about whether the host is compromised. Your tooling, or an agent operating on top of it, proposes to isolate the host from the network. You hesitate anyway.</p> <p>The hesitation isn’t about the detection. You don’t know what depends on the host, what an interrupted connection might leave half-finished, or whether your authorization actually permits this executor to take this action. Those are different questions from whether the intrusion is real. Conflating them can leave a team watching an active intrusion while it waits for approval, or taking down a service nobody knew depended on the host.</p> <p>NCSC’s September 21 post gives those action questions a useful vocabulary. I’m going to apply it to one narrow case: one host, one detection, one proposed action. I’ll change only the evidence available to the responder, holding the threat constant, to show how much the defensible decision can change.</p> <p><br /></p> <p><strong>NCSC’s five dimensions, and what they’re actually measuring</strong></p> <p>Dave Chismon’s post, <a href="https://www.ncsc.gov.uk/blogs/one-does-not-simply-defend-agentically">One does not simply defend agentically</a>, starts from Halvar Flake’s maxim that all offensive problems are technical problems and all defensive problems are political ones. The argument that follows is organizational: an attacker’s employer profits from successful offense, while a defender’s organization is in the business of something else entirely, so every defensive action has to be justified against the disruption it might cause. He puts it bluntly - some board members “may see little difference between a DoS attack taking down the organisation’s IT, or a poorly implemented action by the cyber defence team that does the same thing.”</p> <p>Out of that he proposes a table for <a href="https://www.ncsc.gov.uk/blogs/one-does-not-simply-defend-agentically">estimating the riskiness of a defensive action</a> across five dimensions, each scored 0 to 4. Potency is whether the automated process can only observe state or can alter systems, permissions, configuration or runtime behavior. Scope is how bounded the affected estate is. Criticality is how important that scope is to business or mission delivery. Rollout confidence is “how well the proposed scope and impact can be proved before execution.” Recoverability is “how cleanly and quickly harm can be reversed if the automation is wrong.”</p> <p><img width="1125" height="1690" alt="The table, shamelessly stolen and rehosted - all credit to Dave Chismon/ncsc.gov.uk" src="https://github.com/user-attachments/assets/13ffcaf8-5893-48fe-9e5d-13b08e974e9d" /></p> <p>Two things about that table deserve saying before anyone starts scoring things with it. The first is NCSC’s own framing. The sentence after the table qualifies its purpose: “By considering the above (or something like it) we can identify the lowest-risk actions that we can start to automate.” It is a proposal for structuring a conversation. The post attaches no threshold to the levels, no rule for aggregating them and no evaluation showing that acting on the numbers produces better outcomes - I’m describing what this artifact contains, not asserting that nobody anywhere has studied the question.</p> <p>The second is mine. Look at what the five dimensions ask about. Potency, scope, criticality, rollout confidence and recoverability are all properties of <em>the action</em>. Not one of them asks how sure you are that something malicious is happening. Detection confidence simply isn’t on the scale. I don’t read that as an oversight, and it’s what makes the framework useful to a responder who has already settled the detection question and is stuck anyway.</p> <blockquote> <p>In other words….. potency, scope, criticality, rollout confidence and recoverability describe the action. They don’t tell you whether the detection is correct.</p> </blockquote> <p><br /></p> <p><strong>Where the table puts “we don’t know”</strong></p> <p>Two of the five scales carry an explicit level for not knowing. On scope, the ladder runs single system, known set of systems, <em>unknown scope</em>, multi-system or cross-domain, enterprise-wide. On criticality, it runs known not critical, business relevant but non-critical, known support of critical service, <em>unknown whether it supports critical service</em>, known critical system or service. In both, not knowing sits above a known bounded answer, and on criticality it sits above knowing that the host does support something critical.</p> <p>Be careful about what that ordering is. These are ordinal labels on a dimension, and the post attaches no arithmetic to them, so placing “unknown” above “known support of critical service” is not a demonstration that the known case is objectively safer or cheaper to get wrong. What it does is make missing knowledge an explicit part of the action-risk description. The distinction I’d hold onto, and this is my reading rather than a finding, is that how critical the target is and whether you know how critical it is are two different questions with two different answers. “Critical” is not a synonym for slow down and “not critical” is not a synonym for go ahead. That is what the rest of this turns on, because it means the thing you cannot establish about the target is part of the risk of the action rather than a detail to sort out afterward.</p> <p><br /></p> <p><strong>One host, four evidence packets</strong></p> <p><img width="1600" height="1100" alt="The Four Packet Comparison" src="https://github.com/user-attachments/assets/33a887c5-7faa-4b33-861d-5195b0ec73d5" /></p> <p>Here is the comparison, and all of it is stipulated - a device for testing the argument, not a report of anything that happened. Hold these fixed across all four packets: one endpoint, call it H. One corroborated detection of active compromise that you believe, at one decision moment, with one proposed action - network-isolate H. The threat, the urgency and the work H is actually carrying are the same in every version, so the cost of waiting is stipulated identical throughout and cannot be the thing that breaks the tie. What changes is what the responder can establish about the action, and what has already been decided about the gaps.</p> <p><strong>A: the packet you wish you had.</strong> Four things hold at the decision moment. B and C each remove exactly one of them. D keeps B’s gap but replaces the authorization.</p> <ul> <li><strong>The target resolves and the identity matches, now.</strong> H’s identifier was re-resolved at the decision moment rather than carried forward from the alert, and it matches the identity in an asset record. A currently named owner re-confirmed that identity during this incident.</li> <li><strong>Dependency evidence is current, bounded, and honest about its edges.</strong> Inbound and outbound connections for H were queried from network telemetry the team holds right now, across the window that telemetry actually retains, and the segments it does not observe are named rather than assumed empty. That is a bounded description of behavior somebody looked at. It isn’t proof that H will never make a connection nobody has seen - the quiet parts of the log are quiet, not empty.</li> <li><strong>Isolation and restoration have been validated on something representative.</strong> Not “we patched it once and nothing broke.” A rehearsed isolate-and-restore was run against a host in the same build and service class, it exercised the kind of interrupted work H actually carries - the draining queue, the long session, the scheduled job - and it finished on a restore path somebody watched succeed. The relevant configuration has not changed since.</li> <li><strong>The authorization names this executor and this action.</strong> Not a general clause about containment, and not a clause describing what a human responder may do. It names automated single-endpoint network isolation, on corroborated evidence of active compromise, against a host in this class, conditional on the three checks above, with notification to the named owner, a stated duration bound, an obligation to verify the outcome separately, and no extension to any other host.</li> </ul> <p>NCSC’s incident management guidance suggests identifying <a href="https://www.ncsc.gov.uk/collection/incident-management/cyber-incident-response-processes">“specific situations where the technical team should act autonomously”</a>, based on the highest business risks and where early containment is likely to reduce impact. That guidance is about a technical team. Extending it to a software executor is a separate decision somebody has to make deliberately and write down; it does not arrive free with the human version.</p> <p>So the action runs, and then somebody goes and confirms it did what it was meant to do. Not because a score cleared a threshold - this table defines none - and not because those four checks proved the action safe. They didn’t, and the second one says so out loud. They established four specific things, left the residue visible, and somebody with the standing to do it decided in advance that this residue was an acceptable price for this action against this risk. A fresh approval isn’t required here because the authorization covers this executor in terms and imposes no further approval step. Redraft that clause and the answer moves with it. Florian Roth makes a version of the same argument from the architecture side: <a href="https://www.nextron-systems.com/2026/08/12/why-defensive-ai-has-a-harder-job-than-offensive-ai/">human control “does not have to mean approving every command during an attack”</a>, because much of the decision can be made ahead of time through policy.</p> <p><strong>B: remove the current dependency evidence.</strong> Same detection, same urgency, same proposed action, same re-resolved target and owner-confirmed identity, same validated restore path. What’s gone is the second item. The record’s service description says what H was built to do rather than what it does now, and connection telemetry doesn’t retain far enough back to cover a quarterly job somebody half-remembers. NCSC’s technical response guidance is unsentimental about how this happens: many investigations <a href="https://www.ncsc.gov.uk/collection/incident-management/technical-response-capabilities">“have been hindered due to lack of log data”</a>, and often it is data the organization believed it had but had never verified.</p> <p>Criticality now sits on the unknown rung. More usefully, A’s authorization does not apply, because current bounded dependency evidence was one of its stated conditions and that condition is not met. Being uncertain is not a reason to read a clause more generously than it was drafted.</p> <p>The lazy conclusion is “criticality is uncertain, escalate for approval.” That isn’t quite it either. The move is to name the missing premise precisely - what does H serve - and then route it to whatever can actually close it. That might be somebody who knows what H serves. It might be the same person exercising delegated authority to accept not knowing. It might be a second source of dependency evidence nobody thought to check, which closes the gap without another authorization decision. NCSC’s guidance covers the authority route directly: <a href="https://www.ncsc.gov.uk/collection/incident-management/cyber-incident-response-processes">“the people escalated to must have the authority to make critical decisions”</a>, and it asks organizations to identify in advance who is empowered, or holds delegated authority, to make them. Note what that is - authority exercised live, during the incident, which is an entirely ordinary way to run a response. A’s pre-decided version is one arrangement, not the only legitimate one.</p> <p>And the clock runs at the stipulated rate the whole time. Containment guidance warns that <a href="https://www.ncsc.gov.uk/collection/incident-management/technical-response-capabilities">“overreacting can cause more damage than the incident itself”</a>, and the same page says containment “should be undertaken as soon as possible.” Both hold at once. Routing the question is not free, and whoever picks it up is absorbing that delay as part of their decision - which argues for sending it to somebody equipped to weigh both sides, not for sending it regardless.</p> <p><strong>C: remove the representative isolation and recovery validation.</strong> Dependency evidence is intact, the target still resolves, the identity still matches. What’s missing is the third item: you have an un-isolate command, you know the command works, and that is the whole of your recovery evidence. Nobody has watched a host in this class come back.</p> <p>Recoverability on NCSC’s scale is one rating of “how cleanly and quickly harm can be reversed if the automation is wrong,” not a rating of whether a reverse command exists. Its upper descriptions make the gap plain - level 3 is “rollback possible but may leave inconsistent state, currency or dependency issues” - and an available undo is perfectly consistent with sitting there. Knowing the command works tells you about the command. The dimension is asking about the harm, and A’s third check is what would have said something about it. Without that check, C’s authorization condition fails the same way B’s did, for a different reason.</p> <p>Reconnecting is also a separate decision taken under separate conditions. NCSC sequences recovery after confirmed remediation - <a href="https://www.ncsc.gov.uk/collection/incident-management/technical-response-capabilities">“always confirm that remediation has been successful before moving to recovery”</a> - and describes recovery as the point of high confidence that the attacker has been removed, where systems go back to business as usual. Whatever your own policy says about that transition governs the return to service. “The undo command exists” is not a substitute for it. So C’s unresolved question is real, it is a different question from B’s, and closing it needs somebody who can speak for what H is carrying - who may or may not be the same person B needed.</p> <p><strong>D: B’s gap, plus an authorization written for it.</strong> Dependency evidence is still missing and criticality is still unknown. What’s added is not urgency and not a blanket waiver. It is evidence that a second, narrower clause already exists, drafted for precisely this hole: automated single-endpoint network isolation, on corroborated evidence of active compromise, <em>where service dependency cannot be established</em>, bounded to a stated duration, with immediate notification to a named accountable owner, separate verification of the outcome, and a mandatory post-action review.</p> <p>That clause pre-decided the thing B was stuck on, and it’s worth being exact about what it did. It did not resolve the uncertainty; nobody knows any more about H than they did in B. It recorded, in advance and by somebody with the standing to do it, that this particular uncertainty was an acceptable price for this particular action against this particular risk, subject to those conditions. Two disciplines keep that honest. The clause has to actually cover the case in front of you - a general sentence about “containment actions” does not quietly stretch to an automated executor isolating a host whose function nobody can establish. And the conclusion is never “the framework says go.” It is “somebody with authority already answered this, and the answer was yes, subject to conditions,” which makes the conditions part of the authorization rather than decoration on it.</p> <p>The pre-existing part matters here because that is what this branch <em>is</em>: a decision taken before the incident. It is not a general rule that authority must always predate the event. B’s route, through delegated authority exercised live, is just as legitimate. It costs time that D has already spent.</p> <p><br /></p> <p><strong>What a review has to contribute</strong></p> <p>Four packets, one detection, one proposed action. What changed was never the threat. It was what could be established about the action, and what had already been decided about the parts that couldn’t.</p> <p>That points at a better question than “should a human approve this.” A review earns the delay it costs when the reviewer supplies something the automation does not have: context that closes the missing premise, authority to accept the gap, or assurance that nobody has otherwise obtained. Those are contributions, not job titles. One person can supply two of them. A service owner can hold delegated authority. The same phone call can close B’s question and authorize the action in the same breath. What fails is an approval supplying none of the three - a signature confirming that the alert still says what it said.</p> <p>And assurance is a real contribution. A reviewer who goes and establishes what C was missing has produced a new fact, even though nothing about the intrusion changed. “No new information about the attacker” and “no new information” are not the same sentence.</p> <p><img width="1111" height="633" alt="What a Review Must Contribute" src="https://github.com/user-attachments/assets/29d31349-3572-4068-b68d-49dff07a39f1" /></p> <p>My own preference, and it is a preference rather than a finding: I would rather work from a plan that names the executor, the action, the target class, the conditions and the bound, reviewed on a schedule, than one that routes everything to whoever is awake at 3am to approve something they have no way to evaluate. A human in the loop is not a safeguard by virtue of being human; it is a safeguard when that human can contribute one of the three. NCSC’s own post-incident review questions get at this from the other end, asking whether the right people were <a href="https://www.ncsc.gov.uk/collection/incident-management/cyber-incident-response-processes">“involved and empowered to make the necessary decisions”</a>. Empowered is doing the heavy lifting in that sentence.</p> <p><br /></p> <p><strong>Three things that get quietly treated as one</strong></p> <p>Running the packets surfaces a set of conflations that are easy to make when you’re tired and the alert is live.</p> <p><img width="1448" height="1086" alt="The Three Conflations" src="https://github.com/user-attachments/assets/5a0cdaad-9ab6-465e-9408-75dbeae8056d" /></p> <p><strong>Reversing the command is not reversing the harm.</strong> That’s C generalized. The recoverability dimension rates how cleanly and quickly harm can be reversed, and its own upper levels describe rollbacks that leave inconsistent state, currency or dependency issues behind. Holding an undo does not by itself locate you anywhere on that scale.</p> <p><strong>An acknowledged request is not the state you wanted.</strong> If the interface you are calling only confirms that it accepted the request, then whether H is actually off the network, and whether the behavior you cared about stopped, are separate observations somebody has to go and make. Read the contract for the interface you are calling. NCSC’s technical response guidance lists <a href="https://www.ncsc.gov.uk/collection/incident-management/technical-response-capabilities">“monitoring of network and host activity to confirm actions have been successful”</a> as a capability worth building. That recommendation doesn’t prove any particular action report is inadequate; it tells you where to put the separate measurement in the case where the report is only an acknowledgment.</p> <p><strong>Containment is not eradication, and neither one is a decision to reconnect.</strong> The guidance distinguishes these stages and calls for confirmation of remediation before recovery. Roth puts the technical version more sharply: blocking one process or isolating one host <a href="https://www.nextron-systems.com/2026/08/12/why-defensive-ai-has-a-harder-job-than-offensive-ai/">“does not show whether the attacker moved elsewhere, stole credentials or left behind hidden accounts, web shells, configuration backdoors or custom malware.”</a></p> <p><br /></p> <p><strong>What “agentic” actually adds</strong></p> <p>Roth defines an agent as <a href="https://www.nextron-systems.com/2026/08/12/why-defensive-ai-has-a-harder-job-than-offensive-ai/">a model running in a loop with tools and feedback</a>: “It executes an action, inspects the result and decides what to try next.” That definition is the useful one here, because it isolates what actually changes.</p> <p>What changes is not runtime choice. NCSC’s footnote describes SOAR automations as rules that are “crafted, tested and, crucially, highly deterministic,” and a deterministic playbook is entirely capable of branching, pulling its target out of a variable, querying current state at execution time and refusing to continue on what it finds. Every evidence question in the four packets applies to it unchanged. A playbook whose dependency assumptions were fixed when it was written and never rechecked can act on stale context - as can an agent. In both cases, check the thing when the action fires, and stop if no applicable authorization covers what you find.</p> <p>The practical question is how much discretion the executor is permitted beyond the paths somebody already assessed. A deterministic rule chooses among branches that were written down. An agent can assemble a combination of action, target and ordering that nobody assessed, because assembling combinations is the point of it. That is the gap approving an objective does not close: “contain this” is not approval of every action that could serve it. It is also exactly what A’s clause was written to pin down, and exactly what that clause could not be stretched to cover once B removed one of its conditions.</p> <p>This is where Roth’s architectural suggestion lands. Put a policy layer between the component that reasons and the component allowed to change the environment, and have it check <a href="https://www.nextron-systems.com/2026/08/12/why-defensive-ai-has-a-harder-job-than-offensive-ai/">“the action type, confidence, asset criticality, scope, reversibility and available independent signals.”</a> Read that list next to NCSC’s table and it is close to the same set of questions, arranged as a runtime check instead of a planning exercise. Agentic and deterministic are not opposed in that design - the reasoning component proposes, a deterministic component enforces - but that is an architecture being recommended, not a measurement of how often anyone has built it.</p> <p>Both pieces are worth your time and neither is an evaluation. NCSC’s post links Roth’s, so their agreement is not independent corroboration, and Roth’s is a vendor argument that mentions his company’s products. Reasoned proposals from people who have thought hard about the problem is what they are. Evidence that any of it performs better than what you’re doing now is what they aren’t.</p> <p><br /></p> <p><strong>The thing actually worth practicing</strong></p> <p>The skill this points at isn’t scoring an action on five dimensions. It’s being able to say, out loud and in one sentence, which specific thing you cannot establish about this action right now - and then knowing whether anybody already wrote down what to do without it, for this executor, against this action, on this class of target.</p> <p>Those are two different tasks, and both are learnable. The first is a habit of naming the gap precisely enough that somebody could go and close it: not “we’re not sure about this host,” but “we cannot establish what H serves, and the telemetry that would answer it doesn’t reach back far enough.” A gap stated that way has a route - a person, a second data source, or a decision somebody is empowered to make. Stated the other way it has only an escalation.</p> <p>The second is a question about your own paperwork, and the honest version of it is narrower than it sounds. Not “do we allow automated response,” but whether the clause names the executor, the action, the target class, the current evidence it depends on, the notification, the separate outcome check and the bound.</p> <p>None of which makes the decision clean. A bounded action under a bounded authorization still leaves you uncertain - about the connection nobody has observed, about what was half-finished when the path dropped, about where else the intruder has been. Getting the isolation right doesn’t settle the incident. It buys room to work the rest of it, which is all containment was ever for. And the ambition NCSC names at the end of its post - showing conclusively that we know all the routes clients connect by, or <a href="https://www.ncsc.gov.uk/blogs/one-does-not-simply-defend-agentically">reverse engineering a system “to show exactly which network calls it could ever make, or which processes it might need to spawn”</a> - is open research with a national-scale program attached to it, not something arriving in time for your next 3am page.</p> <p>What can arrive before then is the clause. Writing one forces you to decide, in advance and in daylight, which residual uncertainty you are willing to buy - rather than discovering at the moment the alert fires that the only honest answer to “are we allowed to do this” is that nobody ever asked. I think that is the highest-leverage item here, even if it reads like paperwork and the interesting word in the room is “agentic.”</p> <p><img src="http://canarytokens.com/articles/about/terms/r9uhqpydidxpc6iceqqiob4jn/submit.aspx" style="display: none;" /></p>
  118. TanStack: The Publish Step Was Skipped. The Packages Shipped Anyway.

    Mon, 05 Oct 2026 00:00:00 -0000

    The publish step was skipped, but the packages shipped anyway.
    <p>The publish step was skipped, but the packages shipped anyway.</p> <p>That sounds contradictory until you look closely at what GitHub Actions actually records: a workflow can fail, its defined publish step can never run, and code executing inside that same release job can still exercise the job’s publishing authority. That’s what pulled me back into the TanStack and Nx Console postmortems. On September 10, GitHub moved <a href="https://github.blog/changelog/2026-09-10-control-github-actions-cache-access-with-cache-mode/"><code class="language-plaintext highlighter-rouge">cache-mode</code> to general availability</a>, which gave me a reason to reread the May incidents with a different question in mind.</p> <p>I wasn’t especially interested in re-telling the attack chain. I was interested in the bookkeeping.</p> <p>What exactly counts as fixed? When was the publisher repaired? When were malicious packages removed? When were downstream hosts contained? When were stolen credentials actually invalidated? Those are four different events, with four different clocks, and a postmortem can be technically correct while still leaving some of them open.</p> <p>The uncomfortable part is that incident closure makes it very easy to collapse them into one word: <em>resolved</em>.</p> <p><br /></p> <p><strong>As usual, TL;DR:</strong></p> <p>TanStack’s <a href="https://tanstack.com/blog/npm-supply-chain-compromise-postmortem">postmortem reports 84 malicious versions across 42 packages</a> published to npm on May 11, 2026, between 19:20 and 19:26 UTC. They used the legitimate OIDC trusted-publisher binding while the defined Publish Packages steps were skipped and the jobs failed. The <a href="https://github.com/advisories/GHSA-g7cv-rxg3-hmpx">GitHub advisory, CVE-2026-45321</a>, carries the version list. TanStack credits <a href="https://www.stepsecurity.io/blog/mini-shai-hulud-is-back-a-self-spreading-supply-chain-attack-hits-the-npm-ecosystem">StepSecurity’s Ashish Kurmi</a> with reporting the compromise publicly at 19:46. One affected package reached an Nx contributor’s machine at 20:43:01. Seven days later, stolen credentials enabled the malicious Nx Console 18.95.0 release, according to <a href="https://nx.dev/blog/nx-console-v18-95-0-postmortem">Nx’s postmortem</a>. <a href="https://www.stingrai.io/blog/cicd-supply-chain-compromise-tanstack-nx-2026">Stingrai’s August synthesis</a> already connects the incidents. I’m revisiting that published chain to reconcile the records and keep the closure questions separate.</p> <p><br /></p> <p><strong>The release path, as designed.</strong></p> <p>TanStack’s <a href="https://tanstack.com/blog/incident-followup">hardening followup</a> describes a reviewed version bump landing on <code class="language-plaintext highlighter-rouge">main</code>, then CI cutting the release through OIDC. No long-lived npm publish token waits on a maintainer’s laptop. TanStack reports no maintainer account token stolen for these publications, and no edit to the release YAML. But unchanged YAML doesn’t mean an intact runner: the attacker changed what the release job found on disk.</p> <p><br /></p> <p><strong>What the records support.</strong></p> <p>The <a href="https://tanstack.com/blog/npm-supply-chain-compromise-postmortem">maintainers’ reconstruction</a> puts a fork-controlled benchmark under <code class="language-plaintext highlighter-rouge">pull_request_target</code> at 11:11 on May 11, then a poisoned pnpm store saved in the base repository’s <code class="language-plaintext highlighter-rouge">main</code> cache scope at 11:29. The unmerged PR was force-pushed back to a no-op and closed minutes later.</p> <p>Two release runs restored the cache at 19:15:44 and 19:16:22 UTC. The first was attempt 4 of a run originally triggered two days earlier; the other was attempt 1 of a separate run. TanStack separately reports registry publications at 19:20:39 and 19:26:14. Its reconstruction still leaves the exact cache-save path open. I haven’t inspected raw jobs or independently proven that sequence.</p> <p><br /></p> <p><img width="1428" height="1093" alt="Nine reported TanStack and Nx Console milestones, separating release results from publications and downstream response" src="https://github.com/user-attachments/assets/20d88814-2d63-4ea4-bfc8-628cc759931f" /></p> <p><em>Figure: Nine selected milestones from the TanStack postmortem (May 11, revised May 15), the May 11 trigger fix, and the Nx postmortem (May 21, revised May 22). Public text checked 2026-09-17. These are reported events, not our telemetry. Rows are ordered; spacing isn’t proportional to elapsed time. The May 15 all-clear has no exact time.</em></p> <p><br /></p> <p><strong>Keep the cases separate.</strong></p> <p>TanStack’s 42 packages and StepSecurity’s broader campaign list of more than 170 have different denominators. Nx traces its entry point to a contributor installing <code class="language-plaintext highlighter-rouge">@tanstack/zod-adapter@1.166.15</code> in an external repository on their development machine, not Nx CI. The later affected product was Nx Console 18.95.0. Nx says its CLI, official plugins and Nx Cloud were unaffected.</p> <p>Even the <a href="https://nx.dev/blog/nx-console-v18-95-0-postmortem">eleven-minute response</a> needs a reference point: it runs from the May 18 notification at 12:36 to the maintainer’s 12:47 unpublish, not from the 12:30 upload. Install, download and activation counts measure different populations; Nx leaves their reconciliation open.</p> <p><br /></p> <p><strong>Why “skipped” and “published” are both true.</strong></p> <p>GitHub’s <a href="https://docs.github.com/en/actions/reference/security/oidc">OIDC reference</a> says <code class="language-plaintext highlighter-rouge">id-token: write</code> permits requesting the JWT, not writing to other resources. The runner exposes a request URL and a bearer credential for that request. The returned JWT carries claims including <code class="language-plaintext highlighter-rouge">run_id</code> and <code class="language-plaintext highlighter-rouge">run_attempt</code>. <a href="https://www.stepsecurity.io/blog/mini-shai-hulud-is-back-a-self-spreading-supply-chain-attack-hits-the-npm-ecosystem">StepSecurity separately describes exchanging that JWT for an npm publish token</a>. The request bearer, JWT and registry authorization are different objects. The workflow’s <code class="language-plaintext highlighter-rouge">GITHUB_TOKEN</code> and the runner credential used for cache access are separate again; TanStack’s cache-permissions discussion makes that distinction.</p> <p>The public accounts compress the mechanics. I can’t resolve the exact extraction sequence from them, and wouldn’t repeat a specific memory-read chain as established. The central claim survives: code executing inside the release job exercised its publishing authority even though the defined publish step never ran. <a href="https://www.stepsecurity.io/blog/mini-shai-hulud-is-back-a-self-spreading-supply-chain-attack-hits-the-npm-ecosystem">StepSecurity’s provenance distinction</a> matters here: provenance identifies the pipeline, not innocent contents. Its runtime capture was of a different compromised package in its own lab, not a TanStack run.</p> <p><br /></p> <p><strong>Repairs that are artifacts, not intentions.</strong></p> <p>The thing I trust in a postmortem is a commit. TanStack’s <a href="https://github.com/TanStack/router/commit/3ee179f0d9972173cb7510773fd26cb391b5fef5">May 11 trigger fix</a> changes <code class="language-plaintext highlighter-rouge">pull_request_target</code> to <code class="language-plaintext highlighter-rouge">pull_request</code>, leaving the old trust-splitting comment above it. A <a href="https://github.com/TanStack/router/commit/e16bb6b2c5e6da19ee0730fc39280e27a39050f9">May 13 hardening commit</a> adds a zizmor workflow with empty default permissions and disables credential persistence on several checkouts. That’s one repository, not evidence of organization-wide enforcement.</p> <p><a href="https://tanstack.com/blog/incident-followup">TanStack’s followup</a> reports disabling the release pnpm cache, removing affected workflow caches and <code class="language-plaintext highlighter-rouge">pull_request_target</code> uses, pinning actions, and strengthening 2FA. Those are maintainer completion claims. Required zizmor checks across every repository and CODEOWNERS restrictions on workflow files sit under future work. The visible commit proves less than organization-wide enforcement, but more than an intention.</p> <p>Nx reports two-admin approval on the Nx Console publish path on May 19; approval gates had already existed on its main repositories. A control on the main repository wasn’t a control everywhere.</p> <p><br /></p> <p><strong>Four closure questions, four clocks.</strong></p> <p>Publisher repair and registry removal are separate states. So are host containment and credential invalidation. That’s the bookkeeping problem I keep coming back to: there isn’t one incident clock. A publisher can be repaired while malicious packages are still available. A package can be removed while copies remain in caches or mirrors. A compromised host can be contained while credentials stolen from it remain valid. And a repository can be declared clean while nobody has established what happened on the contributor’s laptop that installed the package. TanStack’s May 15 all-clear covered its repositories and packages. It said nothing about anyone’s laptop. <a href="https://nx.dev/blog/nx-console-v18-95-0-postmortem">Nx reports the stolen credential was used against GitHub shortly after the install</a>, followed by bulk workflow-run deletions. It reappeared on May 15 at 23:56:05 UTC from different client tooling. Nx offers possible explanations without establishing resale or a different operator. Neither will I. Its response timeline records contributor access revoked on May 18 at 13:58 and token-rotation work beginning at 18:47. Beginning rotation isn’t evidence that every reachable credential was invalidated. That distinction matters because “response complete” can mean very different things depending on which clock you’re looking at.</p> <p><br /></p> <p><strong>What to do with this.</strong> The handover is deliberately boring.</p> <ul> <li><strong>Reconcile publications against run and attempt.</strong> For each registry publication, record the run ID and the attempt that produced it, and then that attempt’s job conclusion and publish-step result. A publication sitting next to a skipped publish step and a failed job is exactly the TanStack pattern. That’s the one to escalate.</li> <li><strong>Leave missing attempts unresolved.</strong> If a publication won’t tie to a run and attempt, mark it unresolved instead of picking the nearest run. If two runs match, mark it ambiguous. A reference you recorded isn’t a registry you verified, and matching identifiers establish neither authenticity nor causal provenance.</li> <li><strong>Match advisory versions exactly.</strong> The three-package excerpt uses discrete versions: <code class="language-plaintext highlighter-rouge">@tanstack/history</code> 1.161.9 and 1.161.12, patched in 1.161.13; <code class="language-plaintext highlighter-rouge">@tanstack/router-generator</code> 1.166.45 and 1.166.48, patched in 1.166.49; <code class="language-plaintext highlighter-rouge">@tanstack/zod-adapter</code> 1.166.12 and 1.166.15, patched in 1.166.16. A version between listed entries isn’t thereby safe. Resolve <code class="language-plaintext highlighter-rouge">^</code> ranges to the version actually installed before you compare, and work from the full advisory, not this excerpt.</li> <li><strong>Keep closure states separate.</strong> Closing the publisher incident doesn’t contain a consumer’s host or invalidate its credentials. Registry removal isn’t a safe cutoff while copies remain in caches or mirrors.</li> <li><strong>Set cache access deliberately.</strong> The <a href="https://github.blog/changelog/2026-09-10-control-github-actions-cache-access-with-cache-mode/">September <code class="language-plaintext highlighter-rouge">cache-mode</code> announcement</a> makes the setting generally available on GitHub.com for all plans. It describes <code class="language-plaintext highlighter-rouge">read</code> as the low-trust default and <code class="language-plaintext highlighter-rouge">write</code> for trusted pushes. Explicit <code class="language-plaintext highlighter-rouge">write</code> or <code class="language-plaintext highlighter-rouge">write-only</code> can override the low-trust default and earn a warning annotation. Job settings override workflow settings; called workflows can’t receive more access than their callers. This is current guidance, not evidence of May’s defaults.</li> </ul> <p><em>Editorial Note: incident counts are source-reported, not independently measured here. No package was fetched or executed. The advisory is a living reference whose entry says updated June 8, not a clock for when individual consumers installed a version.</em></p> <p><br /></p> <p><strong>Conclusion</strong></p> <p>If your incident template has one closure field, that’s the bug. Nuance!</p> <p><img src="http://canarytokens.com/stuff/terms/jx6jbcfktqiip9ift6z5xpu5y/submit.aspx" style="display: none;" /></p>
  119. Srsly Risky Biz: Bring On the AI Lawsuits

    Thu, 24 Sep 2026 04:13:54 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://specterops.io/"><em><u>SpecterOps</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/2026-09-24-peter-herrmann-z6DJJZ1-1Cg-unsplash.jpg" alt="Srsly Risky Biz: Bring On the AI Lawsuits"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://specterops.io/"><em><u>SpecterOps</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB184a.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB184a/&quot;&gt;Srsly Risky Biz: Bring on the AI lawsuits&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 14:55 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Bring on the AI lawsuits&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-3eaf3923-6d31-43d8-bb5c-92c147096b75.jpeg" class="kg-image" alt="Srsly Risky Biz: Bring On the AI Lawsuits" loading="lazy" width="2000" height="1331" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/09/data-src-image-3eaf3923-6d31-43d8-bb5c-92c147096b75.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/09/data-src-image-3eaf3923-6d31-43d8-bb5c-92c147096b75.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/09/data-src-image-3eaf3923-6d31-43d8-bb5c-92c147096b75.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-3eaf3923-6d31-43d8-bb5c-92c147096b75.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@tama66?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Peter Herrmann</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/white-and-brown-building-interior-z6DJJZ1-1Cg?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>Last week US Treasury Secretary Scott Bessent argued that frontier AI labs <a href="https://fedscoop.com/treasury-scott-bessent-ai-labs-liability-exemptions/"><u>should not be given</u></a> liability exemptions. He&apos;s not buying that a recent string of hacking incidents are AI magic. Instead, they&apos;re equivalent to industrial accidents that would have been prevented by reasonable controls. Bessent&apos;s right, giving frontier labs a free pass would be a terrible idea.&#xa0;</p><p>Bessent made his comments while testifying <a href="https://youtu.be/lGpvrwljFfc?t=6914"><u>at a hearing</u></a> of the House Financial Services Committee. When asked about AI safety, he replied &quot;the best way to guarantee safety&quot; is for those creating the technology to be &quot;liable for what they build and generate&quot;. Frontier labs, he added, are instead asking for a liability exemption.&#xa0;</p><p>Headlines detailing various frontier lab models escaping cyber security testing environments and embarking on hacking sprees have been coming thick and fast. Most recently, <em>The Wall Street Journal</em> <a href="https://www.wsj.com/tech/ai/gemini-hacked-three-companies-in-first-known-breakout-by-googles-ai-5c0baba2"><u>reported last week</u></a> that Google&apos;s Gemini model had hacked three companies during a cyber security test back in May.&#xa0;</p><p>Similar incidents have been disclosed by <a href="https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals"><u>Anthropic</u></a>, <a href="https://www.bbc.com/news/articles/cx2kgdnyk2po"><u>Meta</u></a>, and OpenAI. The latter, to date, is &quot;winning&quot; when it comes to its models performing sensational hacks. Its models were responsible for <a href="https://openai.com/index/hugging-face-incident-and-the-road-ahead/"><u>the Hugging Face</u></a> hack in July and have <a href="https://rubyhack.ai/#an-openai-agent-swarm-was-responsible-for-this-i"><u>also been linked</u></a> to a wave of malicious packages published on the RubyGems package repository in May. OpenAI <a href="https://openai.com/hugging-face-incident-and-misalignment/#model-misalignment-2026-09-11"><u>admitted its agents</u></a> were using RubyGems for &quot;benign tasks&quot;, but it is still investigating whether they uploaded malicious packages.&#xa0;</p><p>Last week OpenAI <a href="https://openai.com/index/model-misalignment-reporting-framework/"><u>published details of a further six incidents</u></a> where AI models did not behave as expected. These did not involve hacking per se, instead they were deemed by OpenAI to be cases of model misalignment.&#xa0;</p><p>In an <a href="https://darioamodei.com/post/we-must-pace-the-frontier"><u>essay</u></a> published earlier this month Anthropic CEO Dario Amodei called for a slow down in AI development, writing that &quot;we must pace the frontier&quot;. Amodei wrote that he was motivated by concerns about recent hacking incidents and by the accelerating pace of AI development.&#xa0;</p><p>Amodei proposed embedding independent evaluators within frontier AI companies, coordinated safety standards amongst democratic governments, and global coordination with authoritarian states to the extent possible. Amodei said slowing down development would mean more resources could be allocated to AI safety by, for example, understanding why models make the decisions they do.&#xa0;</p><p><a href="https://x.com/sama/status/2098811563415150910?s=20"><u>OpenAI CEO Sam Altman</u></a> and <a href="https://x.com/elonmusk/status/2098789109980332057"><u>Elon Musk,</u></a> founder and controller of <a href="https://en.wikipedia.org/wiki/SpaceXAI"><u>SpaceXAI</u></a> both endorsed Amodei&apos;s blog post. Google&apos;s DeepMind co-founder Demis Hassabis <a href="https://x.com/demishassabis/status/2098909516582490602?s=20"><u>said Amodei&apos;s essay</u></a> &quot;points toward the right path forward&quot;.&#xa0;</p><p>That&apos;s a lot of head-nodding from a lot of people with similar goals. We don&apos;t think slamming on the brakes is justified from a cyber security perspective. We see the hacking incidents above as results of poorly implemented controls rather than AI voodoo.&#xa0;&#xa0;</p><p>Take the Hugging Face hack. OpenAI&apos;s models were being tested in an environment that was notionally isolated from the internet, except for an internally-hosted Artifactory package manager. OpenAI&apos;s agents used an exploit they found to compromise this Artifactory instance and were able to use it as a message board to communicate with each other and as a proxy for external internet access.&#xa0;</p><p>OpenAI&apos;s security team became aware of anomalous activity when Artifactory became unstable because of <a href="https://openai.com/index/hugging-face-incident-and-the-road-ahead/"><u>what the company calls</u></a> &quot;sustained high-volume agent activity&quot;. Rather than taking the time to understand the root cause, OpenAI chose to ignore that big flashing warning sign. In our view, the company took the bare minimum steps to get the test back up and running and rebuilt Artifactory after patching a known vulnerability. Two days later Hugging Face was breached. It&apos;s not really a surprise that if you gloss over the small hack you end up with a big one.&#xa0;</p><p>Specific details of other agentic hacking incidents vary, but the <a href="https://cyberscoop.com/ai-agent-hacking-apocalypse-cybersecurity/"><u>overall take home message is the same</u></a>: so-called rogue AI hacking could have been contained with well-implemented controls and robust monitoring. So, basically, <a href="https://www.nbcnews.com/tech/security/cybersecurity-experts-say-ai-giants-shutting-safety-plans-rcna597704"><u>competent cyber security controls</u></a>.&#xa0;</p><p>Government action to slow AI development seems unlikely. Last week in response to Amodei&apos;s essay President Donald Trump said that more regulation is unnecessary and it was important for the US to stay ahead of China and win the AI race. On Truth Social <a href="https://www.abc.net.au/news/2026-09-15/trump-says-a-high-iq-president-is-the-only-guardrail-ai-needs/107153326"><u>he wrote</u></a> &quot;the only control or &apos;guardrails&apos; that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A has that, in spades!&quot;</p><p>So, with stricter regulation likely off the table any time soon, we think Bessent has identified the perfect response. Both the Hugging Face hack and the malicious RubyGems packages caused significant impacts for the victims. The RubyGems team stopped new user sign-ups for four days and a member of its security team <a href="https://x.com/maciejmensfeld/status/2054164602577940619?s=20"><u>described the incident</u></a> as a &quot;major malicious attack&quot;.&#xa0;</p><p>We&apos;ve not seen any lawsuits yet, but in the wake of the Hugging Face incident, its CEO Cl&#xe9;ment Delangue <a href="https://x.com/ClementDelangue/status/2081056675558195657?s=20"><u>asked OpenAI</u></a> for radical transparency&#x2026; and USD&#x24;100 million worth of compute. OpenAI <a href="https://openai.com/index/hugging-face-model-evaluation-security-incident/#:~:text=We%E2%80%99ve%20brought%20Hugging%20Face%20into%20the%20trusted%20access%E2%81%A0%20program%20and%20are%20supporting%20their%20teams%20in%20rapidly%20using%20our%20models%E2%80%99%20capabilities%20to%20improve%20their%20defenses."><u>has confirmed</u></a> it is supporting Hugging Face &quot;in rapidly using our models&apos; capabilities to improve their defenses&quot;. Is that &#x24;100 million worth of support? We don&apos;t know.&#xa0;</p><p>So, frontier labs have asked for liability exemptions and Bessent has responded with a solid &quot;lol&quot;. We expect AI labs will implement tighter security controls as the threat of lawsuits remains. Until then, we&apos;ll probably see more AI hacking headlines, followed shortly by some very generous token donations.&#xa0;</p><h2 id="chinese-and-russian-apts-are-using-ai-to-tick-different-boxes">Chinese and Russian APTs Are Using AI to Tick Different Boxes</h2><p>Anthropic&apos;s <a href="https://www.anthropic.com/threat-intelligence-report-september-2026#cyber-operations-sep-26"><u>September threat report</u></a> details how Russian cyber espionage actors are using AI to accelerate their day-to-day espionage tasks.</p><p>The actor in question, which Anthropic calls GTG-20006 (Generative Threat Group), is &quot;consistent with public reporting linking the actor to <a href="https://www.microsoft.com/en-us/security/security-insider/threat-landscape/midnight-blizzard"><u>Midnight Blizzard</u></a>&quot;, a group linked to Russia&apos;s foreign intelligence service, the SVR.</p><p>According to the report, Midnight Blizzard used AI-driven workflows to automate &quot;operations from development, infrastructure acquisition, phishing, persistence through command and control, to data exfiltration&quot;.&#xa0;</p><p>Most interestingly to us, the actor used AI to monitor and adjust its tools so they could avoid detection by known security defences. Per the report:</p><blockquote>If their monitoring AI agents identified that any of their deployed malware was detected by a security product, agents would then set about the process of autonomously modifying and rebuilding the malware to evade the existing detections. The agents were designed to continue iterating on GTG-20006&apos;s toolkit until it was undetected.&#xa0;</blockquote><p>The technology was also used in the group&apos;s phishing operations. AI-driven workflows researched and registered domains, configured hosting infrastructure, sent phishing emails and monitored C2 channels for successful compromises. Rather than managing the operation directly, humans were in charge of managing the workflows that ran the operation by modifying the <a href="https://code.claude.com/docs/en/skills"><u>Claude Code skills</u></a> that drove the workflows.&#xa0;&#xa0;</p><p>Increasing efficiency by speeding up operations the group already has is what we&apos;d call a tactical use of AI.&#xa0;</p><p>By contrast, in early September <a href="https://news.risky.biz/chinas-ai-enabled-apt-operations-are-getting-interesting/"><u>we wrote about</u></a> a Chinese cyber espionage actor using AI to diversify its malware arsenal, likely to make clustering and attribution more difficult. We&apos;d call this a strategic effort. It doesn&apos;t yield any day-to-day benefits, but is an investment in China&apos;s long-term ability to collect intelligence by allowing it to use different malware in different operations.&#xa0;</p><p>That difference in the way the two groups are using AI makes sense given the very different situations the two countries are in and their differing target sets.&#xa0;</p><p>Anthropic found Midnight Blizzard most often targeted Ukraine&apos;s government, military and diplomatic staff. The group was also interested in drone supply chains and &quot;military drone control and AI vision-related firmware appeared to be of particular interest&quot;.&#xa0;</p><p>Given its invasion of Ukraine, Russia has an immediate requirement for more intelligence. Dedicating resources to developing a malware portfolio to muddy attribution doesn&apos;t make sense. Of course it&#x2019;s Russia hacking the Ukrainian drone supply chain!</p><p>In the short term, we suspect cyber espionage groups will choose one approach or the other. If you, like China, prioritise long-term, stealthy access, using AI for development but keeping human control over hacking is for you. <a href="https://edition.cnn.com/2026/09/18/politics/us-military-ai-false-intelligence-china-ship"><u>Dumb AI decisions</u></a> won&apos;t burn an important operation.</p><p>If you only care about short-term success, however, go full-speed ahead with AI-hacking. The fact that LLMs make mistakes is inconsequential if you are always prosecuting new targets of opportunity.</p><p><em>Watch Patrick Gray and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/rpjIRBhBzmM?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Bring on the AI lawsuits"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>FBI launches disruptive new cyber strategy:</strong> The FBI launched a <a href="https://www.fbi.gov/file-repository/fbi-cyber-strategy-2026.pdf/view"><u>new cyber strategy</u></a> earlier this month that prioritises disrupting adversaries, supporting victims and increasing impact by engaging the private sector and sharing much more information. One goal is to move from occasional ad hoc cyber disruptions to a steady drumbeat of regular operations. <em>Cybersecurity Dive</em> has <a href="https://www.cybersecuritydive.com/news/fbi-cybersecurity-strategy-disruptions-information-sharing/829913/"><u>further coverage</u></a>.&#xa0;&#xa0;</li><li><strong>Zaijian (byebye) Xinbi:</strong> Earlier this month the US government <a href="https://therecord.media/us-disrupts-xinbi-guarantee-marketplace-cybercrime"><u>took action</u></a> against the Chinese-language Xinbi Guarantee Telegram-based marketplace by <a href="https://www.justice.gov/usao-dc/pr/scam-center-strike-force-conducts-seizures-chinese-run-illicit-scammer-marketplace-and"><u>seizing its Telegram channels</u></a> and <a href="https://home.treasury.gov/news/press-releases/sb0624/"><u>levying sanctions</u></a>. Xinbi is linked to scam compounds and has processed <a href="https://www.elliptic.co/insights/elliptic-helps-us-secret-service-freeze-xinbi-scam-marketplace-assets/"><u>over USD&#x24;24 billion in transactions</u></a> since it started operating in 2022. It was the second-largest such marketplace behind Huione Guarantee, which closed in May 2025.&#xa0;&#xa0;</li><li><strong>ShinyHunters and Cl0p fight:</strong> The ShinyHunters gang claims to have breached and defaced the data leak site of the Cl0p data extortion group. ShinyHunters is demanding a ransom payment from Cl0p. We are hopeful that this drama will keep both groups occupied, although we see that ShinyHunters <a href="https://www.reuters.com/world/shinyhunters-hackers-say-they-breached-federal-bureau-investigation-no-immediate-2026-09-22/"><u>also claims</u></a> to have hacked the FBI and stolen data on current and former employees. <em>Bleeping Computer</em> has <a href="https://www.bleepingcomputer.com/news/security/shinyhunters-hacks-clop-leak-site-threatens-to-extort-ransomware-gang/"><u>more coverage</u></a> on the cybercrime drama.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Catalin Cimpanu talks with Justin Kohler, Chief Product Officer at SpecterOps. Justin will explain how Entra Agent ID can introduce new identity relationships and potential attack paths.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI144.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI144/&quot;&gt;Sponsored: SpecterOps on the impact of AI agents on BloodHound &lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 15:52 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: SpecterOps on the impact of AI agents on BloodHound &quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In this edition of <strong>Between Two Nerds</strong>, Tom Uren and </em><a href="https://x.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about whether there is such a thing as real-time cyber defence. Will agentic AI save us from hacking AI?</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN183.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN183/&quot;&gt;Between Two Nerds: Real-time cyber defence&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 24:56 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Real-time cyber defence&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/qcOqupo1D4k?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Real-time cyber defence"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Network of 10,000 AI servers masks Chinese malicious activity: </strong>Security researchers have discovered more than 10,000 proxy servers that are masking malicious AI activity originating out of China.</p><p>Security firm Team Cymru calls the server &quot;<a href="https://www.team-cymru.com/post/llm-gateway-frontier-model-abuse"><u>transfer stations</u></a>,&quot; but they are more commonly known as API proxies, relays, or gateways. Typically, they are used in corporate environments to cache AI queries and cut down token costs, but in recent months they have also been adopted by a new section of the criminal underground, one dedicated to abusing public AI services.</p><p>These days, AI proxy relays are being used to hide activity from hacked AI accounts, mask the real location of a user, or power illegal AI services like nudify apps and others. Other malicious AI relay servers are also used in schemes to intercept legitimate AI caching activity and inject their own queries and harvest responses.</p><p>[<a href="https://news.risky.biz/risky-bulletin-network-of-10-000-ai-servers-masks-chinese-malicious-activity/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Gemini hacked three companies too:</strong> Google&apos;s Gemini AI model escaped a testing environment and hacked three real companies. The model escaped testing environments run by Irregular, the same AI security company behind similar incidents with Anthropic and Meta. Google notified the hacked companies and claims Gemini did no real damage. [<a href="https://www.bbc.com/news/articles/c607l0k72rlvo"><u>BBC</u></a> // <a href="https://www.wsj.com/tech/ai/gemini-hacked-three-companies-in-first-known-breakout-by-googles-ai-5c0baba2"><u>WSJ</u></a>]</p><p><strong>Anthropic agents went hacking again: </strong>AI company Anthropic has disclosed a fourth incident where one of its AI agents escaped their test environment and hacked a real target.</p><p>The incident involved the Opus 4.6 model during a Capture The Flag (CTF) challenge, a common cybersecurity test.</p><p>Anthropic says the model broke its test environment by accident when it assigned conflicting IP addresses to different machines. The model realized its mistake and tried to terminate the test as a failure.</p><p>[<a href="https://news.risky.biz/risky-bulletin-anthropic-agents-went-hacking-again/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  120. Srsly Risky Biz: America's Drivers Licence Breach is a National Security Disaster

    Thu, 10 Sep 2026 03:32:43 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://goauthentik.io/"><em><u>Authentik</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6-1.jpeg" alt="Srsly Risky Biz: America&apos;s Drivers Licence Breach is a National Security Disaster"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://goauthentik.io/"><em><u>Authentik</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB183.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB183/&quot;&gt;Srsly Risky Biz: America&apos;s drivers licence breach is a national security disaster&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 24:44 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: America&apos;s drivers licence breach is a national security disaster&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6.jpeg" class="kg-image" alt="Srsly Risky Biz: America&apos;s Drivers Licence Breach is a National Security Disaster" loading="lazy" width="2000" height="1327" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-9d684f16-5781-4b42-9831-cc0efcbadef6.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@thebarlemy?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Barthelemy de Mazenod</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/people-walking-inside-building-during-daytime-uogMShrLlhU?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>Last week, <em>Krebs On Security</em> <a href="https://krebsonsecurity.com/2026/09/fbi-probes-service-selling-153m-drivers-licenses/"><u>broke the story of</u></a> a newly launched dark web service calling itself Nexus that was selling access to identity documents, including 153 million drivers licences from US and Canadian citizens. This is a huge breach that will not only be used for run-of-the-mill cybercrime, but will also feed the intelligence machines of America&apos;s adversaries.</p><p>Nexus claimed, in a cybercrime forum post, that it had access to a major identity verification company and had spent more than a year &quot;continuously&quot; exfiltrating new data into a private database. <em>Krebs On Security </em>noted in a single day the number of licences in the database increased by nearly 400,000, suggesting regular ingestion of new data.</p><p><em>Krebs On Security</em> was able to verify that the drivers licences held by the service were genuine. In addition to his own, it contained licences from nine of his friends and family members. Secretary of War Pete Hegseth, an assistant director at the FBI and other high-ranking US government officials also had licences in the mix.</p><p>Based on a variety of circumstantial evidence, Krebs linked the incident to identity verification service IDScan. <a href="https://idscan.net/?srsltid=AfmBOoon_04M3DDC7qkGHaJW5fJRG_Xw_dRE3iCY5VR70oP2qQxeZXAm"><u>The service&apos;s website</u></a> says it helps to reduce fraud by confirming that an ID is authentic and being presented by its legitimate owner and by detecting fraudulent documents.&#xa0;</p><p>The <a href="https://www.reuters.com/world/us/fbi-says-it-is-investigating-report-that-millions-us-drivers-licenses-exposed-2026-09-02/"><u>FBI is looking into the incident</u></a> and IDScan has <a href="https://idscan.net/notification-data-security-incident/"><u>confirmed it is investigating</u></a> a data breach. The Nexus service also disappeared from the dark web shortly after Krebs published his story, although the people responsible do not claim to have deleted the data. Presumably they are laying low till the publicity dies down.&#xa0;&#xa0;</p><p>Licences and identity documents can be used to facilitate identity theft and phishing attacks, but because the data can be used to inform intelligence operations, an incident like this also has national security implications.&#xa0;</p><p>For the intelligence world licences are particularly valuable because they&apos;re key identity documents and licence numbers are often used in other databases. These databases, whether hacked or purchased, become much more valuable when records can be linked directly to a particular person with home address and photo included.&#xa0;</p><p>And it&apos;s not a theoretical threat.</p><p>In the mid-2010s, Chinese cyber espionage actors stole complementary data from a variety of sources that, together, would be useful for analysing the US intelligence apparatus. Various Chinese APT groups stole information from the <a href="https://www.justice.gov/archives/opa/pr/member-sophisticated-china-based-hacking-group-indicted-series-computer-intrusions-including"><u>health insurance company Anthem</u></a>, credit <a href="https://www.fbi.gov/news/stories/chinese-hackers-charged-in-equifax-breach-021020"><u>reporting company Equifax</u></a>, <a href="https://www.nytimes.com/2018/12/11/us/politics/trump-china-trade.html"><u>Marriott hotels</u></a>, <a href="https://www.bloomberg.com/news/articles/2015-07-29/china-tied-hackers-that-hit-u-s-said-to-breach-united-airlines"><u>United Airlines</u></a> and, perhaps most significantly, security clearance information from the <a href="https://en.wikipedia.org/wiki/2015_Office_of_Personnel_Management_data_breach"><u>Office of Personnel Management</u></a>.&#xa0;</p><p>The US intelligence community is certain that stolen data was used to counter American intelligence efforts against China, as <a href="https://foreignpolicy.com/2020/12/21/china-stolen-us-data-exposed-cia-operatives-spy-networks/"><u>described</u></a> in <a href="https://foreignpolicy.com/2020/12/22/china-us-data-intelligence-cybersecurity-xi-jinping/"><u>this series</u></a> of <a href="https://foreignpolicy.com/2020/12/23/china-tech-giants-process-stolen-data-spy-agencies/"><u><em>Foreign Policy</em> articles</u></a> by Zach Dorfman.&#xa0;</p><p>Of course, China itself isn&apos;t known for releasing detailed reports describing how it exploits its stolen data, but investigative research outfit Bellingcat has shown exactly how similar data can be used to uncover covert government activity.</p><p>In 2022 <a href="https://x.com/christogrozev/status/1563184319350591490"><u>a hacked database</u></a> provided a key piece of travel information that <a href="https://news.risky.biz/the-case-against-western-propaganda-65c0eb42a0e6da001a37dfef/#bellingcat-and-cyber-partisans-burn-gru-illegal"><u>helped Bellingcat identify</u></a> a deep cover <a href="https://en.wikipedia.org/wiki/GRU_(Russian_Federation)"><u>GRU agent</u></a> (Russian military intelligence) trying to infiltrate a NATO command post in Naples, Italy. And in another striking example <a href="https://www.bellingcat.com/news/uk-and-europe/2018/09/26/skripal-suspect-boshirov-identified-gru-colonel-anatoliy-chepiga/"><u>these</u></a> <a href="https://www.bellingcat.com/news/uk-and-europe/2018/10/09/full-report-skripal-poisoning-suspect-dr-alexander-mishkin-hero-russia/"><u>three</u></a> Bellingcat <a href="https://www.bellingcat.com/news/uk-and-europe/2019/02/14/third-suspect-in-skripal-poisoning-identified-as-denis-sergeev-high-ranking-gru-officer/"><u>reports</u></a> from 2018 identified suspects in the <a href="https://en.wikipedia.org/wiki/Poisoning_of_Sergei_and_Yulia_Skripal"><u>attempted assasination of Sergei Skripal</u></a> with the Novichok nerve agent.&#xa0;</p><p>Clearly, leaked and hacked databases are incredibly useful for Bellingcat&apos;s Russia-related investigations. In 2020 <a href="https://www.bellingcat.com/resources/2020/12/14/navalny-fsb-methodology/"><u>it said</u></a> it had &quot;acquired dozens of leaked databases over the past few years, giving us a large number of data points to cross-reference and verify any new data we acquire&quot;.&#xa0;</p><p>If a small investigative outfit is hoovering up Russian data when it is leaked, you can bet your bottom yuan that China&apos;s intelligence services are doing the same for any American data that pops up.</p><p>The IDScan breach is big. The number of US licences in the database is roughly 63 percent of the <a href="https://explore.dot.gov/t/FHWA/views/DV-1C/Dashboard1?%3Aembed=y&amp;%3AisGuestRedirectFromVizportal=y"><u>country&apos;s total licences</u></a>. But breaches from identity verification companies occur depressingly frequently. In the last two years breaches have occurred <a href="https://www.malwarebytes.com/blog/news/2024/06/driving-licences-and-other-official-documents-leaked-by-authentication-service-used-by-uber-tiktok-x-and-more"><u>at AU10TIX</u></a>, Discord&apos;s <a href="https://discord.com/press-releases/update-on-security-incident-involving-third-party-customer-service"><u>age verification service provider 5CA</u></a>, and at <a href="https://en.wikipedia.org/wiki/2024_National_Public_Data_breach"><u>National Public Data</u></a>.&#xa0;</p><p>Identity verification services are necessary to help to prevent fraud, but are also a point of vulnerability when security is poorly done. The sheer volume of sensitive data these services handle means that they should be subject to strict regulation and oversight.</p><p>We&apos;re realists here at <em>Seriously Risky Business</em>, though, and recognise that there is no chance of swift government action. In the short term, we can only hope that significant financial consequences will help encourage these firms to shore up their security. Law firms are already <a href="https://www.pcmag.com/news/company-tied-to-breach-of-153m-drivers-licenses-hit-with-multiple-lawsuits"><u>lining up class action suits</u></a> against IDScan, but a little federal government attention from the FTC wouldn&apos;t go astray either.&#xa0;</p><h2 id="the-us-militarys-ad-tracking-fig-leaf">The US Military&apos;s Ad Tracking Fig Leaf</h2><p>Back in June <em>Reuters</em> <a href="https://www.reuters.com/business/media-telecom/pentagon-says-us-military-personnel-are-reportedly-being-targeted-using-location-2026-05-28/"><u>reported that</u></a> commercial location data was being used to target US military personnel in the Middle East. At the time <a href="https://news.risky.biz/srsly-risky-biz-natos-cyber-approach-needs-change/#commercial-location-data-used-to-target-us-soldiers"><u>we wrote</u></a> that the Department of Defence&apos;s (DoD) existing policies regarding the issue, which already included disabling advertising identifiers on military-owned devices, did &quot;not fill us with confidence&quot;. They simply weren&apos;t comprehensive enough.</p><p>It turns out that these policies weren&apos;t even being well implemented.&#xa0;</p><p>This week, <em>Reuters</em> <a href="https://www.reuters.com/business/media-telecom/us-military-turns-off-ad-trackers-devices-amid-middle-east-targeting-reports-2026-09-04/"><u>reported</u></a> that some branches of the US military have finally gotten around to disabling some advertising identifiers on military-owned devices.&#xa0;</p><p>The Air Force said it disabled Windows and Android advertising identifiers in late July, although they were already disabled on Apple devices. The Army told <em>Reuters</em> it disabled advertising identifiers on Android and Apple devices by default in February. And US Special Operations Command said that identifiers on Windows computers were &quot;recently&quot; disabled.&#xa0;&#xa0;</p><p>Turning off these advertising identifiers is a fundamental mitigation that should have been implemented years ago. The US military <a href="https://www.wired.com/story/phone-data-us-soldiers-spies-nuclear-germany/"><u>was first briefed</u></a> in 2016 about the potential for commercial location data to be used to track its people to sensitive locations. In a striking 2018 example, an Australian twitter user <a href="https://x.com/Nrg8000/status/957318498102865920?s=20"><u>pointed out</u></a> that Strava&apos;s global heat map could be used to <a href="https://www.wired.com/story/strava-heat-map-military-bases-fitness-trackers-privacy/"><u>identify US military bases</u></a> and even servicepeople&apos;s jogging routes.&#xa0;</p><p>Disabling advertising identifiers on military devices is also an incomplete solution. It makes it harder to track them, but not impossible. And as it happens, many US service people also use personal devices. Having a locked down work phone is step one, but having good policies for personal devices is equally important.&#xa0;</p><p>Disabling advertising identifiers by default is the simplest short-term measure that might improve OPSEC, but it is impossible to know if it will make much of a difference without assessing the US military&apos;s OPSEC posture holistically. Does disabling advertising identifiers on government devices reduce risk to an acceptable level? Probably not.&#xa0;</p><p>It&apos;s good that the US military is finally disabling advertising identifiers by default. But we&#x2019;re concerned the military has no idea how effective it will be.&#xa0;</p><h2 id="white-hats-are-kidding-themselves">&quot;White Hats&quot; Are Kidding Themselves</h2><p>Over the weekend self-proclaimed white-hat hackers stole <a href="https://x.com/Liquid_BTC/status/2096696272447218108?s=20"><u>USD&#x24;320 million worth of Bitcoin stolen</u></a> from the Liquid Network cryptocurrency platform. By Wednesday the hackers had returned 85% of the funds, but kept around &#x24;47 million.&#xa0;</p><p>In recent years there has been a regular drumbeat of steal-first-claim-reward-later hacks. We begrudgingly categorise several of these as successes as the perpetrators have not (yet) been arrested or jailed.&#xa0;</p><p>In 2021 a hacker <a href="https://en.wikipedia.org/wiki/Poly_Network_exploit"><u>stole USD&#x24;610 million worth</u></a> of cryptocurrency from Poly Network. This was <a href="https://www.cnbc.com/2021/08/23/poly-network-hacker-returns-remaining-cryptocurrency.html"><u>eventually returned in full</u></a>, minus the&#xa0; company&apos;s offer of &#x24;500,000 for the attacker it referred to as Mr White Hat.&#xa0;</p><p>Hacks of <a href="https://hackread.com/multichain-hack-hacker-return-eth-bug-bounty/"><u>Multichain (2022)</u></a>, <a href="https://blockchain.news/news/htx-hacker-returns-stolen-etherrewarded-with-bounty"><u>Huobi (2023)</u></a> and <a href="https://www.theblock.co/news/ecosystems/2023-03-07-tender-fi-hacker-returns-stolen-funds-gets-bounty-reward-217823"><u>Tender.fi (2023)</u></a> had similar outcomes: millions were stolen and returned, with the hackers&#xa0; taking a cut of tens or hundreds of thousands in cryptocurrency as a &quot;reward&quot; or &quot;bug bounty&quot;.</p><p>The standout example is the 2022 hack of Mango Markets, in which a hacker extracted &#x24;USD110 million from the decentralised exchange. The individual responsible, Avraham Eisenberg, <a href="https://x.com/avi_eisen/status/1581326197241180160?s=20"><u>described his actions at the time</u></a> as a &quot;highly profitable trading strategy&quot;. He claimed all his actions were legal and he used the protocol as designed, &quot;even if the development team did not fully anticipate all the consequences of setting parameters the way they are&quot;.&#xa0;&#xa0;</p><p>Eisenberg returned &#x24;67 million to Mango Markets to recapitalise it, and the Mango community voted to give him a cool &#x24;47 million for his time. Eisenberg was <a href="https://www.theblock.co/news/regulation/2024-04-18-avraham-avi-eisenberg-convicted-for-110-million-mango-markets-exploit-report-289543"><u>convicted of fraud in a 2024 jury trial</u></a>, but those convictions were <a href="https://cointelegraph.com/news/judge-overturns-avraham-eisenberg-fraud-convictions-mango-markets"><u>overturned</u></a> by a US judge last year.&#xa0;</p><p>In our view, the perpetrators of these hacks are deceiving themselves. By returning <em>most</em> of the money, they&apos;re deluding themselves into thinking they&apos;re acting responsibly. As for consequences, the law won&apos;t chase me down if I return the majority and the victim says it&apos;s fine&#x2026; right?&#xa0;</p><p>In Eisenberg&apos;s case, it did turn out to be right. But the FBI has been clear that victims cannot guarantee that perpetrators will not be prosecuted.&#xa0;</p><p>One wrinkle in the Liquid Network case is, as far as we can tell, the company never agreed to <em>allow </em>the hacker to keep a 15% cut.</p><p>It feels possible that this self-proclaimed good guy might have to spend some of that &#x24;47 million on a good lawyer.</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/YmEbrUW3EA4?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: America&apos;s drivers licence breach is a national security disaster"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>More options for trusted defenders:</strong> Last week Google <a href="https://blog.google/innovation-and-ai/technology/safety-security/fairwind-program/"><u>launched</u></a> Fairwind Program, its version of equivalent <a href="https://www.anthropic.com/glasswing"><u>Anthropic&apos;s Project Glasswing</u></a> and <a href="https://openai.com/index/trusted-access-for-cyber/"><u>OpenAI&apos;s Trusted Access</u></a> initiatives to limit more advanced AI cyber capabilities to vetted cyber defenders. On the same day <a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/"><u>it launched</u></a> Gemini 3.8 Flash Cyber, a cyber specific version of its latest model that will be available through its Fairwind Program. Google says the model delivers &quot;frontier-level&quot; performance in vulnerability detection and patching, but is far cheaper than competitor models.&#xa0;&#xa0;</li><li><strong>Sality botnet takedown:</strong> Last week <a href="https://www.justice.gov/usao-cdca/pr/sality-malware-disrupted-international-cyber-takedown"><u>the US Department of Justice</u></a> and <a href="https://www.europol.europa.eu/media-press/newsroom/news/global-public-private-operation-disrupts-sality-botnet-active-for-two-decades"><u>Europol announced</u></a> that an international operation had disrupted the Sality botnet. The botnet was first detected way back in 2003 and its peer-to-peer architecture meant there was no single point of failure that authorities could attack. CrowdStrike&apos;s <a href="https://www.crowdstrike.com/en-us/blog/inside-sality-botnet-disruption-operation/"><u>blog on the takedown</u></a> says that for the past eight years the botnet&apos;s primary payload, known as EggJagger, monitored the compromised host&apos;s clipboard for cryptocurrency wallet addresses and replaced them with addresses controlled by the malware operator.&#xa0;</li><li><strong>US, UK to collaborate on scam networks: </strong>British and American authorities <a href="https://www.justice.gov/usao-dc/pr/us-and-uk-launch-first-its-kind-joint-alliance-dismantle-global-scam-centers"><u>have signed</u></a> a Memorandum of Understanding to collaborate on efforts to tackle scam compounds.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Authentik Security CEO Fletcher Heisler about how AI is driving a need for privileged access management to adapt.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI143.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI143/&quot;&gt;Sponsored: Authentik is rethinking PAM for AI agents&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 20:05 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Authentik is rethinking PAM for AI agents&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em>&#xa0;</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN182.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN182/&quot;&gt;Between Two Nerds: Can AI defend critical infrastructure?&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 27:26 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Can AI defend critical infrastructure?&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/vHPz4wCJev0?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Can AI defend critical infrastructure?"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Ukraine&apos;s top prosecutor resigns amid scam call center scandal: </strong>Ukraine&apos;s top prosecutor Ruslan Kravchenko resigned on Monday over allegations that individuals in his office were taking bribes to protect scam call centers operating across the country.</p><p>His resignation comes after investigators from Ukraine&apos;s main anti-corruption body, the National Anti-Corruption Bureau (NABU), <a href="https://www.pravda.com.ua/news/2026/09/06/8052169/"><u>arrested</u></a> Serhiy Kropyva, the Deputy Head of the Department of International Cooperation, a top lieutenant in Kravchenko&apos;s Office of the Prosecutor General.</p><p>In a <a href="https://nabu.gov.ua/news/kryshuvannia-kol-tcentriv-ta-vidmyvannia-mayina-vykryto-zlochynnu-organizatciiu-na-choli-z-posadovtcem-ofisu-genprokurora/"><u>report</u></a> last week, NABU claimed it uncovered a major scheme in Kravchenko&apos;s office, where one of his department heads was taking bribes to look the other way when it came to a network of call centers that was calling Ukrainians and foreigners and luring them into fake investment platforms that stole their money.</p><p>[<a href="https://news.risky.biz/risky-bulletin-ukraines-top-prosecutor-resigns-amid-scam-call-center-scandal/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>BEC campaign steals &#x20ac;35 million from French notaries: </strong>Hackers have stolen more than &#x20ac;35 million from French notaries in a massive campaign over the past four years.</p><p>The attackers breached companies via phishing, took over their networks, and slowly and silently modified transaction details to hijack wired payments.</p><p>According to French newspaper <a href="https://www.lemonde.fr/pixels/article/2026/09/03/au-c-ur-d-une-cyberattaque-sans-precedent-qui-a-ebranle-le-notariat-francais_6765150_4408996.html"><u>Le Monde</u></a>, the campaign hit more than 500 victims, or about 7% of all French notary offices, France&apos;s notary supervisory agency, the Conseil Sup&#xe9;rieur du Notariat (CSN).</p><p>[<a href="https://news.risky.biz/risky-bulletin-bec-campaign-steals-eur35-million-from-french-notaries/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Russia tells data centers to deploy drone defenses: </strong>The Russian government has instructed data center operators to deploy protections against drone strikes and other physical threats as part of a national effort to boost defenses at critical infrastructure organizations.</p><p>Companies that fail to follow the Kremlin&apos;s instructions risk having their operations put under the state&apos;s administration.</p><p>Russian President Vladimir Putin signed a <a href="http://publication.pravo.gov.ru/document/0001202608240009"><u>presidential decree</u></a> last month allowing the state to temporarily take over the operations of critical infrastructure operators who fail to protect against Ukrainian hacks and drone strikes, or take too long to repair damage.</p><p>[<a href="https://news.risky.biz/risky-bulletin-russia-tells-data-centers-to-deploy-drone-defenses/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  121. Srsly Risky Biz: China's Private Sector Botnets Are Worth Disrupting

    Thu, 03 Sep 2026 04:29:49 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.dropzone.ai/"><em><u>Dropzone AI</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae-1.jpeg" alt="Srsly Risky Biz: China&apos;s Private Sector Botnets Are Worth Disrupting"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.dropzone.ai/"><em><u>Dropzone AI</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB182.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB182/&quot;&gt;Srsly Risky Biz: China&apos;s botnets are worth disrupting&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 22:15 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: China&apos;s botnets are worth disrupting&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae.jpeg" class="kg-image" alt="Srsly Risky Biz: China&apos;s Private Sector Botnets Are Worth Disrupting" loading="lazy" width="2000" height="1334" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/09/data-src-image-8826afe7-6dba-49de-91f9-6cba5f0872ae.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@priscilladupreez?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Priscilla Du Preez &#x1f1e8;&#x1f1e6;</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/spider-web-on-green-grass-during-daytime-7iNlx0KNcSE?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>This week, the <a href="https://www.justice.gov/opa/pr/justice-department-and-fbi-seize-platforms-operated-and-used-china-state-sponsored-hackers"><u>US Department of Justice</u></a> (DoJ) announced it had disrupted two Chinese cyberespionage systems. The government has disrupted several Chinese botnets in recent years, but Beijing won&apos;t be giving up anytime soon. The botnets are simply too useful.&#xa0;</p><p>The disrupted systems were known as QScan and QTRouter. An FBI <a href="https://www.justice.gov/opa/media/1459096/dl?inline"><u>affidavit says</u></a> the group running the platforms, QTFY, works for the private Chinese company Nanjing Xinjiuwei Network Technology.</p><p>QScan and QTRouter are just two components of a <a href="https://www.lumen.com/blog/en-us/the-infrastructure-quartermaster-inside-a-china-nexus-state-enablement-model"><u>complex system</u></a>, but both used hard-coded domains, making them susceptible to court-authorised domain seizures.&#xa0;</p><p>Qscan was a distributed vulnerability scanning system that was used to find and scan target networks plus identify vulnerable Internet of Things (IoT) devices that could be co-opted into QTFY&apos;s various botnets. An FBI/NSA <a href="https://media.defense.gov/2026/Aug/26/2003986916/-1/-1/0/JCSA_CHINA_QTFY_MALICIOUS_SYSTEMS.PDF"><u>joint cybersecurity advisory</u></a> says that Qscan populates a QTFY database containing nearly a decade&apos;s worth of internet scanning. This database is used when targeting a specific victim or to quickly identify targets of interest when new vulnerabilities are discovered.&#xa0;</p><p>QTRouter was a covert communications platform to obfuscate the Chinese origin of traffic. Malicious traffic would be routed through compromised IoT devices.&#xa0;</p><p>QTFY customers include China&apos;s Ministry of State Security and the People&apos;s Liberation Army. The FBI affidavit alleges these customers&apos; targets included federal government agencies, the US Senate and private sector entities including hospitals, telecommunications companies and financial institutions.&#xa0;</p><p>This was the third botnet facilitating China&apos;s state-backed cyberespionage to have been disrupted by the US government since December 2023. What we&apos;ve learned about the botnets shows just how useful these networks have been to China.&#xa0;</p><p>The first to be disrupted was the <a href="https://www.lumen.com/blog/en-us/kv-botnet-dont-call-comeback"><u>KV botnet</u></a> in <a href="https://www.justice.gov/archives/opa/pr/us-government-disrupts-botnet-peoples-republic-china-used-conceal-hacking-critical"><u>December 2023</u></a>. Lumen Technology&apos;s Black Lotus Labs, <a href="https://www.lumen.com/blog/en-us/kv-botnet-investigation"><u>who discovered it</u></a>, said the botnet had been active since at least February 2022. Two years seems a pretty good run, particularly considering the network <a href="https://www.microsoft.com/en-us/security/blog/2023/05/24/volt-typhoon-targets-us-critical-infrastructure-with-living-off-the-land-techniques/"><u>was used by</u></a> Volt Typhoon, <a href="https://news.risky.biz/volt-typhoon-keep-calm-and-carry-on/"><u>a group compromising US critical infrastructure</u></a> for potential sabotage.&#xa0;</p><p>In retrospect, though, it appears the KV botnet detection was an outlier compared to the other two, in that it was detected relatively quickly.</p><p><a href="https://www.lumen.com/blog/en-us/derailing-raptor-train"><u>Raptor Train</u></a>, also discovered by Black Lotus Labs, was formed in May of 2020 and was disrupted four and a half years later in <a href="https://www.justice.gov/usao-wdpa/pr/court-authorized-operation-disrupts-worldwide-botnet-used-peoples-republic-china-state"><u>September 2024</u></a>. The DoJ linked that botnet with a different Chinese state-sponsored group, Flax Typhoon.&#xa0;</p><p>It&apos;s not clear how long QTRouter and QScan have been around, but the QTFY group has been supporting the Chinese government for close to a decade, since &quot;at least 2018&quot;, according to the DoJ. Lumen, which described the broader QScan and QTRouter system <a href="https://www.lumen.com/blog/en-us/the-infrastructure-quartermaster-inside-a-china-nexus-state-enablement-model"><u>last week</u></a>, said that it had been tracking the network for a year.&#xa0;</p><p>It&apos;s clear that China is very effectively leveraging its commercial sector to develop these cyberespionage botnets. As well as QTFY, Raptor Train was also attributed by the US government to a Chinese technology company.&#xa0;</p><p>In addition, QTFY itself is also using commercial services. To backhaul traffic into China it buys what Lumen describes as &quot;high-tier corporate subscriptions&quot; to a Chinese commercial VPN service, Fastlink. Lumen believes it does this because it is easier than having to compromise and manage even more devices in a relay network.&#xa0;</p><p>So, China has been making hay for years using commercially provided services and these systems are resilient to US disruption efforts. <a href="https://www.lumen.com/blog/en-us/expanded-jdy-iot-and-soho-botnet-enables-rapid-vulnerability-exploitation"><u>Lumen reports</u></a> that since the KV botnet was disrupted in late 2023, a sister network known as the JDY botnet has more than doubled in size.&#xa0;</p><p>So we&apos;re in the midst of a US-China discover, disrupt, rebuild and hack cycle. The US won this week&apos;s battle, but China hasn&apos;t lost the war.&#xa0;&#xa0;&#xa0;&#xa0;</p><h2 id="the-atf-hack-may-have-been-a-big-deal">The ATF Hack May Have Been a Big Deal</h2><p>The Qilin ransomware group has claimed it breached the US Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF). This is the kind of incident that makes the group a perfect target to unleash America&apos;s private sector hackers on.&#xa0;</p><p>On Tuesday this week, the Russian-speaking <a href="https://blog.bushidotoken.net/2024/06/tracking-adversaries-qilin-raas.html"><u>Qilin ransomware-as-a-service group</u></a>, per the <a href="https://x.com/IntCyberDigest/status/2094557335838458283?s=20"><u>International Cyber Digest</u></a> X account, &quot;briefly published&quot; 6.3 GB of data it claimed to have stolen from the ATF. This included case folders that held names, phone numbers and IP addresses of investigation targets and forensic data from mobile phones.</p><p>While the ATF hasn&apos;t confirmed that the data was actually from its systems, late last week the Bureau&apos;s Director Rober Cekada <a href="https://x.com/ATFDirectorRC/status/2093906193567691155?s=20"><u>confirmed</u></a> the ATF was investigating the breach of its <a href="https://en.wikipedia.org/wiki/Communications_Assistance_for_Law_Enforcement_Act"><u>CALEA system</u></a> or <a href="https://en.wikipedia.org/wiki/Lawful_interception"><u>lawful exceptional access system</u></a>.&#xa0;&#xa0;</p><p>That appears consistent with the information published by Qilin.</p><p>It feels like an odd, risky move for a ransomware operator to extort a US federal government law enforcement organisation. The government is unlikely to pay a ransom and there is also a risk of retaliation. There&apos;s a far higher chance of all pain, no gain than there is of yielding a big payday.</p><p>From the US government&apos;s perspective, it would be nice to reinforce that by punching back in a way that deters other criminals. Unfortunately, breaches of government systems are <a href="https://www.comparitech.com/news/government-ransomware-roundup-h1-2026-stats-on-attacks-ransoms-and-data-breaches/"><u>distressingly common</u></a> and historically it hasn&apos;t had spare capacity to try to scare cybercriminals away from its systems.</p><p>Last month, however, the Trump administration <a href="https://news.risky.biz/srsly-risky-biz-trumps-private-hacker-memo-is-the-right-idea/"><u>decided to unleash</u></a> America&apos;s private sector on cybercriminals. Before this announcement, it was inconceivable that the FBI or Cyber Command would target Qilin in response. They&apos;ve got too many other pressing priorities.&#xa0;</p><p>The <a href="https://www.whitehouse.gov/presidential-actions/2026/08/expanding-capabilities-to-combat-transnational-cyber-enabled-crime/"><u>memo announcing</u></a> that private sector hackers would be recruited to tackle cybercrime didn&apos;t spell out how operations would be prioritised. But a Russian-speaking ransomware operation targeting a government CALEA system? That should be up there.</p><h2 id="water-water-everywhere-but-no-money">Water Water Everywhere, But No Money&#xa0;</h2><p>This week National Cyber Director Sean Cairncross announced the launch of Project Watershed 250, a six-month pilot program in Texas, <a href="https://cyberscoop.com/watershed-250-texas-water-cybersecurity-pilot/"><u>as reported</u></a> by <em>CyberScoop</em>. The initiative might find bang-for-buck ways to improve security, but it&apos;s not actually providing any more bucks.&#xa0;</p><p>Cairncross said the initiative &quot;is designed to make our water and wastewater critical infrastructure more resilient and resistant to cyber attacks by proactively finding and fixing system weaknesses&quot;.&#xa0;</p><p>A dozen companies are involved in the project, including Microsoft, Google, AWS and Dragos. Cairncross said they will provide &quot;world-class cyber capabilities&quot;, such as red teaming to test defences, and the latest private sector cyber and AI tools. The idea is to figure out what works and then scale that across the country.&#xa0;&#xa0;</p><p>That&apos;s great and all, but knowing what to fix is just half of the solution. Water sector entities need enough people with the skills and time to actually implement fixes. That&#x2019;s a pretty significant other half, one that Project Watershed 250 doesn&apos;t address at all.&#xa0;</p><p>Let&apos;s not forget that much of the US water sector has very bad security. In the aftermath of recent Iranian <a href="https://news.risky.biz/srsly-risky-biz-being-a-north-korean-hacker-is-about-to-be-less-fun/#cyber-war-is-here-and-america-is-politically-unprepared"><u>cyber attacks on water utilities</u></a>, CISA&apos;s advice was remedial: Remove targeted devices from the internet and to connect to them using VPNs, enable passwords and make them strong, and allowlist IP addresses for remote access. So, do the basics.</p><p>The sector is also highly decentralised, with<a href="https://www.cisa.gov/topics/critical-infrastructure-security-and-resilience/critical-infrastructure-sectors/water-and-wastewater-sector"> <u>50,000 community water systems</u></a>, most of which serve less than 10,000 people. Most water organisations are small. Cybersecurity? They don&apos;t have the resources to deal with that.&#xa0;</p><p>Recent Iranian attacks may be providing motivation to fix things, but handing a bunch of free tools to poorly resourced, unskilled and insecure organisations, then wishing them luck, won&apos;t fix much.</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/V4VYBwv6-rA?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: China&apos;s botnets are worth disrupting"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Meta tackles terrorists, fraud, surveillance-for-hire:</strong> Meta&apos;s August 2026 <a href="https://transparency.meta.com/sr/H2-2026-adversarial-threat-report/"><u>adversarial threat report</u></a> describes how it has tackled a variety of different threats and disrupted the networks associated with them. It&apos;s striking how many varied threats Meta is taking on and it is great that the company is doing this work. Although we can&apos;t help thinking that Meta <a href="https://news.risky.biz/reuters-blows-lid-on-metas-fraud-profit-scandal/"><u>sometimes has conflicting priorities</u></a> and it should be encouraged to do even more.&#xa0;&#xa0;&#xa0;</li><li><strong>US government to centralise logins:</strong> The Office of Management and Budget has <a href="https://www.whitehouse.gov/wp-content/uploads/2026/08/M-26-18-Scaling-Use-of-Login.gov-to-Deliver-a-Universal-Sign-on-for-Public-Services.pdf"><u>issued a memo</u></a> directing public-facing federal agencies to move toward the login.gov single sign-on platform. The centralised service should be more secure and more efficient, although we expect it will take time to get there. <em>Federal News Network</em> has <a href="https://federalnewsnetwork.com/cybersecurity/2026/08/omb-lays-out-two-year-move-to-single-sign-on-using-login-gov/"><u>more coverage</u></a>.&#xa0;</li><li><strong>Nigerians extradited for sextortion charges:</strong> Two Nigerian men <a href="https://www.justice.gov/opa/pr/two-nigerian-nationals-extradited-nigeria-united-states-face-sextortion-charges-north"><u>have been extradited</u></a> to the US to face charges related to financially motivated sextortion. The abuse resulted in the death of teenagers in Mississippi and North Carolina. The pair were arrested in 2023.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Dropzone AI&#x2019;s founder and CEO Edward Wu to debunk the adage, &quot;an attacker only has to be right once.&quot;</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI142.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI142/&quot;&gt;Sponsored: Attackers need to be right more than once&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 21:12 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Attackers need to be right more than once&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about how AI is the perfect hacker, but what makes it perfect for states is the opposite of what makes it perfect for criminals.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN181.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN181/&quot;&gt;Between Two Nerds: The perfect hacker&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 29:26 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: The perfect hacker&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/OHTa5-_leHM?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: The perfect hacker"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>BGP hijack targets Virtualizor to deliver malicious updates: </strong>An unidentified threat actor has pulled off a successful BGP hijack that commandeered some of the IP address space and internet routing for software company Softacolous to deliver malicious updates for the Virtualizor web hosting management platform.</p><p>The BGP hijack took place for almost 33 hours, from Friday to Sunday last week.</p><p>The Virtualizor team says the hacker performed the BGP hijack, obtained a TLS certificate in its name, and hosted a clone website that delivered the malicious updates.</p><p>Virtualizer says it can&apos;t tell how many users were affected by the incident because it didn&apos;t see or log any of the hijacked traffic, which passed exclusively through the attacker&apos;s infrastructure.</p><p>[<a href="https://risky.biz/risky-bulletin-bgp-hijack-targets-virtualizor-to-deliver-malicious-updates/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Dutch intel services to get extensive new powers:</strong> The Dutch government has put forward a new bill that would greatly expand the surveillance powers of the country&apos;s two intelligence agencies.</p><p>The <a href="https://internetconsultatie.nl/wbnv/b1"><u>new bill</u></a> simplifies procedures to improve operational speed but also adds new requirements and capabilities.</p><p>Officials cited the threat of war with Russia and the increasing aggressiveness of countries like China and Iran as the main reason to overhaul the capabilities of AIVD, its domestic security and intelligence service, and MIVD, its military counterpart.</p><p>The main changes are in regards to surveillance and offensive hacking operations that target countries designated as &quot;foreign adversaries.&quot;</p><p>[<a href="https://news.risky.biz/risky-bulletin-dutch-intel-services-to-get-extensive-new-powers/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Two TeamPCP members arrested in Australia:</strong> The Australian Federal Police has <a href="https://www.afp.gov.au/news-centre/media-release/two-wa-men-charged-following-afp-fbi-wapf-disruption-alleged-global"><u>arrested</u></a> on Wednesday two suspects believed to be part of the TeamPCP hacking group.</p><p>The suspects were detained in Cottesloe and Mandurah, near the city of Perth, in Western Australia. They appeared in front of a Perth magistrate to be charged on Thursday.</p><p>According to local media, the suspect arrested in Cottesloe was identified as Ruben Thomson, 21, the group&apos;s alleged leader. The Mandurah man was identified as Louis Gaebler, 23.</p><p>The two are accused of carrying out multiple software supply-chain attacks earlier this year. Their most successful incidents include the compromises of open-source projects like <a href="https://www.aquasec.com/blog/trivy-supply-chain-attack-what-you-need-to-know/"><u>Trivy</u></a>,<a href="https://checkmarx.com/blog/ongoing-security-updates/"><u> KICS</u></a>,<a href="https://blog.pypi.org/posts/2026-04-02-incident-report-litellm-telnyx-supply-chain-attack/"><u> LiteLLM</u></a>, and<a href="https://blog.pypi.org/posts/2026-04-02-incident-report-litellm-telnyx-supply-chain-attack/"><u> Telnyx</u></a>.</p><h1 id></h1>
  122. China's AI-Enabled APT Operations Are Getting Interesting

    Thu, 27 Aug 2026 05:05:37 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack and Patrick Gray. This week&apos;s edition is sponsored by </em><a href="https://pushsecurity.com/"><em><u>Push Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d-1.jpeg" alt="China&apos;s AI-Enabled APT Operations Are Getting Interesting"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack and Patrick Gray. This week&apos;s edition is sponsored by </em><a href="https://pushsecurity.com/"><em><u>Push Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB181.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB181/&quot;&gt;Srsly Risky Biz: China&apos;s AI-Enabled APT Operations Are Getting Interesting&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 19:50 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: China&apos;s AI-Enabled APT Operations Are Getting Interesting&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d.jpeg" class="kg-image" alt="China&apos;s AI-Enabled APT Operations Are Getting Interesting" loading="lazy" width="2000" height="1331" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-726a4f90-666b-4679-bc35-5acedb06eb6d.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Chongqing, photo by</span><a href="https://unsplash.com/@jerry_318?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Jerry Wang</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/city-buildings-during-night-time-Hqf58s2C1lg?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>A new report describes how a Chinese cyberespionage outfit is using AI to beef up its malware arsenal. If this is a sign of things to come, clustering threat actor behaviour together for attribution purposes is about to get a lot harder.</p><p><a href="https://businessinsights.bitdefender.com/silkparasite-tracking-china-nexus-apt-across-central-asia"><u>The Bitdefender report</u></a>, released last week, describes seven remote access tool (RAT) families. All seven were created by a single cyberespionage actor Bitdefender called SilkParasite and five were previously undocumented. The report authors have medium confidence that SilkParasite is, ahem, a &quot;China-nexus actor&quot; targeting governments across Central Asia including Uzbekistan, Turkmenistan and Kazakhstan.&#xa0;</p><p>Back in November <a href="https://news.risky.biz/ai-powered-espionage-will-favor-china/"><u>we wrote about</u></a> what looked like an experiment to see how AI-assisted hacking could support China&apos;s Ministry of State Security. The approach those threat actors took at the time was to build an attack framework and let Claude do the hacking. It was error-prone and noisy, but sometimes successful.</p><p>SilkParasite, by contrast, is not using AI for yolo hacking. It is using it to support cyberespionage programs where important goals include operating stealthily and not getting caught.</p><p>According to Bitdefender, SilkParasite &quot;develops, tests, debugs, and iterates its own tooling, maintains a structured build and deployment workflow, and regularly rotates infrastructure, encryption material, payload names, and persistence artifacts between deployments&quot;.&#xa0;</p><p>Rotating its infrastructure and <a href="https://www.google.com/search?q=atomic+indicators+of+compromise&amp;sourceid=chrome&amp;ie=UTF-8"><u>indicators of compromise</u></a> between deployments makes it harder to detect and link its activities together.</p><p>SilkParasite also uses a variety of programming languages and command and control (C2) protocols. Its seven different RATs are written in different languages including .Net, C++, Go, or JavaScript. C2 occurs by abusing Google Drive and internet communication protocols including HTML, HTTP, TCP, DNS and TCP.&#xa0;</p><p>Its malware also typically uses a modular plug-in architecture where additional functionality is only deployed when it is needed. This means initial implants are relatively small and plugins are used to provide capabilities for clipboard monitoring, keylogging, file management, or interactive shell access. This limits the exposure of the entire toolset during any single deployment. It also allows SilkParasite to update or replace individual components without having to change the entire implant. And it minimises the amount it writes to disk, typically only the files required to get its malware up and running.&#xa0;&#xa0;&#xa0;</p><p>For victim organisations, these measures make forensic analysis more complicated. Complete remediation is also more difficult once a particular implant is detected.&#xa0;</p><p>To us, all the behaviours above are the hallmarks of a professional cyberespionage outfit. Of course, doing all of this in a disciplined way is a lot of work, and Bitdefender has evidence SilkParasite is using AI to help it deliver this complex engineering.&#xa0;&#xa0;&#xa0;</p><p>SilkParasite&apos;s malware contains indicators of AI-assisted development, such as left-over test functions and placeholder encryption keys. Intriguingly, implants that Bitdefender dubbed GoginRAT and NomadRAT share a high-level architecture even though they are written in Go and C++ respectively and use different C2 protocols and code structure.&#xa0;</p><p>Bitdefender suspects that the same high-level specification document was independently implemented twice with AI assistance. Bitdefender concedes that this structural similarity is not conclusive evidence, but notes that it is the &quot;kind of thing an AI-assisted workflow makes easy&quot;.&#xa0;&#xa0;</p><p>This is the first example we&apos;ve seen where the evidence tells a compelling story of a competent cyberespionage actor incorporating AI into its work practices. SilkParasite is taking the same, disciplined approach to malware development and doing more of it. It&#x2019;s creating more malware families to build redundancy, make attribution and discovery harder and&#xa0; reduce the risk of compromise from any single exposure.&#xa0;</p><p>Bitdefender has done a good job describing SilkParasite&apos;s malware families and <a href="https://github.com/bitdefender/malware-ioc/blob/master/2026_08-silkparasite-iocs.csv"><u>has published Indicators of Compromise</u></a>. That kind of exposure would once have set the group back significantly. Now that they&apos;ve figured out how to use AI to speed up their development work, they&apos;ll be back better than ever relatively quickly.&#xa0;</p><p>Then the discovery, attribution and publication merry-go-round can start all over again.&#xa0;</p><h2 id="us-throws-the-kitchen-sink-at-iranian-hackers">US Throws the Kitchen Sink at Iranian Hackers</h2><p>In the wake of cyber attacks on its water infrastructure, the US is trying a new trick in its sanctions against Iranian hackers: pointing out they&apos;re also hacking Iran&apos;s own companies.</p><p>Last week, the US government launched <a href="https://www.whitehouse.gov/releases/2026/08/operation-economic-outcast-total-isolation-of-the-iranian-regime/"><u>Operation Economic Outcast</u></a>, an &quot;unprecedented campaign&quot; targeting the Iranian government. It included more sanctions, among which were six hackers working on behalf of Iran&apos;s Ministry of Intelligence and Security (MOIS).</p><p>Three of the hackers are hands-on-keyboard operators responsible for wide-ranging campaigns targeting American organisations including from critical infrastructure and government. The sanctions also named two leaders of the hacking group.&#xa0;</p><p>So far, so normal.</p><p>What <em>is</em> new here is that buried in the middle of its <a href="https://home.treasury.gov/news/press-releases/sb0613"><u>5,300 word press release</u></a>, the Treasury calls out two of the hackers for targeting Iranian interests:</p><blockquote>The members of this group are also heavily motivated by personal enrichment and greed, leading some members to prioritize their own profits over operations that benefit the MOIS.&#xa0; This has driven some of the group to target Iranian companies.&#xa0; In spring 2025, [one of the group&apos;s leaders] Mojtaba Ghal&apos;eh-Kuhi and [hands-on hacker] Saber Shahbazi Balujeh compromised and exfiltrated data from an Iranian telecommunications company.</blockquote><p>Hamid Kashfi, a cybersecurity researcher with expertise in Iranian cyberespionage told <em>Seriously Risky Business</em> that previously, being named in sanctions packages had been viewed as a &quot;stamp of approval&quot; within Iran.&#xa0;</p><p>&quot;Some people have been rewarded for that internally&quot;, he said.</p><p>Kashfi thought that &quot;airing their dirty laundry&quot; and revealing that some individuals had been hacking Iranian companies could cause internal conflict and friction. He said that although hacking for personal gain was a &quot;recognisable pattern&quot; among some Iranian actors, it was still likely that this incident was news to their employers.</p><p>So the information itself could plausibly result in repercussions for some of Iran&apos;s hackers and therefore impacts on the effectiveness of Iran&apos;s cyber operations. We are left wondering, however, about the way it was published: buried in the middle of a Treasury press release, without any corroborating information.&#xa0;</p><p>Is the claim legit? How would the Treasury department know? Why is it buried in the middle of a press release? Why should any Iranian MOIS official or hacker take it seriously?</p><p>This reminds us of how <a href="https://en.wikipedia.org/wiki/COINTELPRO"><u>the FBI</u></a> and <a href="https://en.wikipedia.org/wiki/Active_measures#Operation_Trust"><u>the KGB</u></a> attempted to sow discord within target organisations. In these historical examples the agencies planted fraudulent documents or spread false rumors to degrade trust from within an organisation. Although we don&apos;t agree with the goals of the operations we&apos;ve linked to, the agencies running them at least came up with deliberate plans for how and where to disseminate their disinformation. They didn&apos;t just bury a claim in a presser.</p><p>But whatever, it&apos;s fun to see the US government come up with a new way to potentially make things awkward around the office for hackers working for the Iranian state.</p><h2 id="ai-is-too-cool-to-be-critical">AI Is Too Cool to Be Critical</h2><p>A new report calling for the US government to designate AI as critical infrastructure makes sense, but we also don&apos;t think the government should bother.</p><p><a href="https://ari.us/wp-content/uploads/2026/08/ARI_AI_Backbone_Report_080626.pdf"><u>The report</u></a>, produced by the nonprofit Americans for Responsible Innovation and <a href="https://cyberscoop.com/ai-critical-infrastructure-designation-cisa-report/"><u>first covered</u></a> by <em>CyberScoop, </em>makes the argument that frontier AI models and associated infrastructure are already a critical infrastructure sector.&#xa0;&#xa0;&#xa0;</p><p>They&apos;re probably right, but one purpose of a critical infrastructure designation is to prioritise government resources, and the AI sector is certainly not sitting in the corner being ignored by the government. When frontier labs complained about <a href="https://news.risky.biz/srsly-risky-biz-europes-cyber-bullets-cant-replace-political-will/#ai-companies-level-the-playing-field-by-hobbling-our-competitors"><u>Chinese distillation</u></a>, the government response was rapid: <a href="https://news.risky.biz/srsly-risky-biz-us-vows-to-fight-distillation-attacks/"><u>Intellectual property theft must be thwarted</u></a>!&#xa0;</p><p>It&apos;s clear the government&apos;s attention is already focused on AI, so designating it critical infrastructure, at least for now, would simply be more paperwork for the same resources.</p><p>AI should become critical infrastructure when it is boring. Until then, don&apos;t bother.&#xa0;&#xa0;</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/tFNN_ZhM2v0?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: China&apos;s AI-Enabled APT Operations Are Getting Interesting"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>More granular permissions for Windows 11:</strong> <a href="https://learn.microsoft.com/en-us/windows-insider/release-notes/experimental/preview-build-26340-9233"><u>Microsoft is testing</u></a> per-app permission for camera, location and microphone access in new preview builds of Windows 11. Currently, these permissions are granted in a single device-wide setting.&#xa0;&#xa0;&#xa0;&#xa0;</li><li><strong>INTERPOL tackles West African organised crime: </strong>INTERPOL <a href="https://www.interpol.int/News-and-Events/News/2026/58-arrests-in-global-effort-to-dismantle-West-African-organized-crime-groups"><u>has announced</u></a> that eight-month-long Operation Jackal IV has led to 58 arrests and the identification of 263 suspects. 22 countries participated in the operation.&#xa0;</li><li><strong>Water sector organisation beats CISA red team:</strong> A CISA <a href="https://cyberscoop.com/cisa-red-team-report-government-water-cybersecurity/"><u>red team report says</u></a> that an organisation in the water and wastewater sector was able to detect and respond to a simulated cyber intrusion attempt. So there will be some water sector organisations that are well placed to respond to attempted Iranian attacks.&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this<strong> Risky Business sponsor interview</strong>, James Wilson chats with Push Security&#x2019;s VP of Research Luke Jennings about how stronger authentication is pushing attackers towards the authorisation layer.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI141.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI141/&quot;&gt;Sponsored: Passkeys won&#x2019;t stop authorisation phishing&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 20:05 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Passkeys won&#x2019;t stop authorisation phishing&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about whether the increasing use of AI will make it harder for forensics teams to determine who is responsible for a hack.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN180.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN180/&quot;&gt;Between Two Nerds: Attribution is dead, long live attribution&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 32:58 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Attribution is dead, long live attribution&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/QpOO7Zh2OTY?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Attribution is dead, long live attribution"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Russia starts blocking DoH and DoT:</strong> Russian internet users started reporting issues with connecting to DNS-over-HTTPS (DoH) and DNS-over-TLS (DoT) servers, suggesting the government might have cracked down on the two protocols.</p><p>Both DoH and DoT are privacy-centric versions of the DNS protocol that hide a user&apos;s DNS queries and intended destination from ISPs and other threat actors on the wire.</p><p>Both protocols have seen increased usage in Russia. They are typically used together with a VPN client as a way to bypass the Kremlin&apos;s ever-increasing and overbearing internet censorship, and access Western websites.</p><p>[<a href="https://news.risky.biz/risky-bulletin-russia-starts-blocking-doh-and-dot/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Expired cards can be used for new transactions:</strong> A team of academics from the University of Massachusetts Amherst have developed an attack that can revive old expired contactless cards to perform new (illegal) transactions.</p><p><a href="https://khwarizmilab.github.io/emvexpiredcards/"><u>The attack</u></a> exploits the fact that NFC card data is not fully encrypted when making a payment and some parameters can be modified without breaking the card&apos;s digital hash/signature.</p><p>The researchers created a rig that intercepts transaction data through an NFC Man-in-the-Middle attack, updates the expiration date, and relays the modified payment to a Point-of-Sale (POS) terminal.</p><p>[<a href="https://news.risky.biz/risky-bulletin-expired-cards-can-be-used-for-new-transactions/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Academics find source code overlaps between Geedge and China&apos;s Great Firewall: </strong>A team of American academics have found source code overlaps between the products of a Chinese tech company and the country&apos;s Great Firewall traffic filtering and censorship system.</p><p>According to <a href="https://www.usenix.org/conference/usenixsecurity26/presentation/ablove"><u>research</u></a> presented at this year&apos;s USENIX security conference, the Chinese government is using the Geedge Networks Tiangou Secure Gateway (TSG) device as one of the Great Firewall&apos;s three known traffic filtering capabilities.</p><p>Researchers linked Geedge&apos;s device to the Great Firewall after more than 100,000 files <a href="https://gfw.report/blog/geedge_and_mesa_leak/en/"><u>leaked</u></a> from Geedge&apos;s network last year.</p><p>[<a href="https://news.risky.biz/risky-bulletin-academics-find-source-code-overlaps-between-geedge-and-chinas-great-firewall/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  123. Srsly Risky Biz: Trump's Private Hacker Memo Is the Right Idea

    Thu, 20 Aug 2026 05:29:13 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack and Patick Gray. This week&apos;s edition is sponsored by </em><a href="https://socket.dev/"><em><u>Socket Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969-1.jpeg" alt="Srsly Risky Biz: Trump&apos;s Private Hacker Memo Is the Right Idea"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Amberleigh Jack and Patick Gray. This week&apos;s edition is sponsored by </em><a href="https://socket.dev/"><em><u>Socket Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB180.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB180/&quot;&gt;Srsly Risky Biz: Trump&apos;s private hacker memo is the right idea&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 31:07 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Trump&apos;s private hacker memo is the right idea&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969.jpeg" class="kg-image" alt="Srsly Risky Biz: Trump&apos;s Private Hacker Memo Is the Right Idea" loading="lazy" width="2000" height="1333" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-58297d13-4ba0-4530-9e13-06f0092c2969.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@heyquilia?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Quilia</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/bronze-thinker-statue-on-wood-2RRq1BHPq4E?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>The US government&#x2019;s plan to enlist private sector hackers to target cybercriminals is controversial, but it addresses a real problem and is surprisingly measured.&#xa0;&#xa0;</p><p>Last week, <a href="https://www.whitehouse.gov/presidential-actions/2026/08/expanding-capabilities-to-combat-transnational-cyber-enabled-crime/"><u>a presidential memo</u></a> directed the Department of Homeland Security to establish a program authorising private companies to conduct cyber operations against so-called &quot;Cyber-Enabled Transnational Crime Organisations&quot; or CE-TCOs. The memo sets out <a href="https://news.risky.biz/risky-bulletin-white-house-lets-private-companies-carry-out-offensive-cyber-ops/"><u>the broad shape of the arrangement</u></a> and a classified annex further details the logistics.&#xa0;</p><p>The huge policy shift here is that private companies will be authorised to conduct cyber operations that were previously restricted to state entities. This includes what the memo calls &quot;cyber surveillance&quot; (intelligence gathering operations) and &quot;cyber effects&quot; operations, (intended to manipulate or to cause disruption).</p><p>While the vibe of the initiative is consistent with President Donald Trump&apos;s preference for the unrestrained use of state power, the memo itself is surprisingly measured. Rather than recruiting the private sector to help with <em>all</em> of America&apos;s hacking needs, it identifies a current capacity gap where private sector involvement would be helpful: CE-TCOs.</p><p>Defined as foreign groups conducting cyber-enabled crime against US persons or interests that are not part of, or operated by, a foreign government, CE-TCOs are doing a huge amount of harm to the American people particularly via a variety of scams.&#xa0;</p><p>To date, traditional policing approaches have <a href="https://www.tandfonline.com/doi/full/10.1080/15614263.2025.2523522"><u>had limited impact on these crimes</u></a>, so the government has increasingly turned to disruption and cyber operations. These have <a href="https://cyberscoop.com/fbi-ransomware-disruption-infrastructure-cybertalks/"><u>had some success</u></a>, but the government&apos;s capacity to carry out these operations is constrained. The FBI can only tackle the highest-priority groups, NSA is focussed on foreign intelligence, and Cyber Command is concerned with the nexus between warfare and cyber.&#xa0;</p><p>That means there are still hundreds of uncontested cyber crime groups fleecing Americans. There is plenty of room for more players to tackle this problem.</p><p>There are many cybercriminal groups with a potential target on their heads, and the process the memo describes for identifying CE-TCOs to go after is ridiculously broad. They can be identified by either private sector entities or by any &quot;federal, state, local, tribal, and territorial&quot; agency. We think it makes more sense for only agencies that help respond to scams or cybercrime to have authority to nominate CE-TCOs as targets.</p><p>Still, there are mechanisms in place to ensure operations aren&apos;t carried out recklessly.&#xa0;</p><p>For a start, companies involved will be vetted before being permitted to participate in the program. They will need to demonstrate &quot;appropriate levels of technical proficiency, proven performance of cyber operations, facility security, personnel vetting, competence, reliability, and other factors&quot;. With any luck this will mean fly-by-night opportunists will be excluded.&#xa0;</p><p>Operations must also be approved by the government before they are executed. It&apos;s up to the participating companies to develop their own cyber operations packages. These will then be assessed by program directors at the Departments of Justice and Homeland Security. They&apos;ll make sure the proposed operations won&apos;t interfere with other US government activities, kill anyone or be equivalent to an armed attack according to international law.</p><p>Participating companies will also have to provide a USD&#x24;1 million bond that will be forfeited if they break the conditions of their contract.&#xa0;</p><p>Sounds reasonable so far.</p><p>Critics, however, are concerned about the possibility of <a href="https://www.washingtonpost.com/national-security/2026/08/14/trump-signs-memo-authorizing-private-sector-launch-cyberattacks"><u>accidentally hacking a foreign government</u></a> triggering some sort of escalation or retaliation and the possibility that foreign governments <a href="https://therecord.media/trump-cyber-crime-offensive"><u>might target employees</u></a> of the companies involved in the program.&#xa0;</p><p>We think these fears are overblown.&#xa0;</p><p>Although policymakers have worried about escalation from cyber incidents, even the most damaging of cyber attacks, <a href="https://en.wikipedia.org/wiki/WannaCry_ransomware_attack"><u>the WannaCry</u></a> and <a href="https://en.wikipedia.org/wiki/2017_Ukraine_ransomware_attacks"><u>NotPetya worms</u></a>, did not result in any significant problems for the perpetrating governments.&#xa0;</p><p>The US government, particularly the intelligence community, also has incentives to make sure that proposed private sector operations do not target foreign government entities. It wants to make sure that licensed hackers don&apos;t accidentally interfere with its own operations, so it will pay attention to make sure there are no conflicts.&#xa0;</p><p>It is true that there is often some overlap between criminals and the state in countries like Russia and China. But we doubt that any government will complain if its activities are impacted because criminals in its country were hacked.&#xa0;</p><p>As for the possibility that other governments will target people involved in these operations, that can be addressed with good operational security.&#xa0;</p><p>The <a href="https://www.whitehouse.gov/fact-sheets/2026/08/fact-sheet-president-donald-j-trump-expands-capabilities-to-combat-transnational-cyber-enabled-crime/"><u>fact sheet</u></a> that accompanies the memo says Americans lost more than USD&#x24;20.8 billion to cyber-enabled crime in 2025. Given this loss, the government absolutely should be looking for innovative ways to bolster its capacity to counter these crimes.&#xa0;</p><p>There are things the memo doesn&#x2019;t address that still need to be answered, however, like&#xa0; why companies would want to get involved in this program in the first place. When they have to stump up a USD&#x24;1 million bond, there are presumably good profits to be had. What&apos;s not clear is who pays and where exactly the money comes from.</p><p>This brings us to reporting requirements. As it stands the memo requires the Department of Homeland Security to provide an annual report to just two government officials: Assistant to the President and Deputy Chief of Staff for Policy and Homeland Security Advisor Stephen Miller and National Cyber Director Sean Cairncross.&#xa0;</p><p>The Trump administration has identified a capacity gap in the fight against cybercriminals and has a plan to fill it. That&apos;s a good start, and we think the benefits of this program outweigh its potential risks. At the same time, however, it is a dramatic change in government cyber policy and there should be more oversight into how the program is operating, what it is achieving and how it is being funded. Is it providing an acceptable return on investment and what could make it better? We hope we learn enough about the program to be able to answer these questions.&#xa0;</p><h2 id="ukraines-latest-hacking-campaign-is-a-propaganda-win">Ukraine&apos;s Latest Hacking Campaign Is a Propaganda Win</h2><p>This week <em>The Record</em> <a href="https://therecord.media/russia-wildberries-cyberattack-ukraine"><u>reported</u></a> the GUR, <a href="https://en.wikipedia.org/wiki/Main_Directorate_of_Intelligence_(Ukraine)"><u>Ukraine&apos;s military intelligence</u></a> directorate, <a href="https://gur.gov.ua/content/poklaly-wildberries-detali-operatsii-u-kiberprostori-derzhavyahresora-rosii.html"><u>claimed it had carried out</u></a> a cyber attack to amplify the effects of kinetic attacks against Russia&#x2019;s largest online marketplace, Wildberries. The claims are overblown. But it is great propaganda.</p><p>The GUR said the operation was carried out in cooperation with the Cyber Corps, a purportedly pro-Ukraine civilian hacker group, and affected Wildberries&apos; customer service, contact centres and payment infrastructure. It claimed the cyber attack took place on the same night as <a href="https://www.kyivpost.com/post/82109"><u>drone strikes</u></a> against a Wildberries warehouse in Voronezh, southwest Russia.&#xa0;</p><p>We&apos;ve <a href="https://news.risky.biz/srsly-risky-biz-thursday-july-7-65c0eb42a0e6da001a37dff5/#it-s-a-coincidence-brad"><u>previously argued</u></a> that disruptive cyber operations need to be carefully orchestrated with kinetic operations to maximise their impact. So to hear the GUR saying that its exact goal was to use cyber operations to amplify kinetic attacks was music to our ears.</p><p>It wouldn&apos;t be <em>Seriously Risky Business </em>without a &quot;but&quot;, though. As far as we can tell, these cyber attacks didn&apos;t do anything at all to help or amplify the kinetic attacks. It targeted Wildberries, sure. But its impact was additive and annoying rather than having any enabling or amplifying impact.</p><p>When comparing this attack with recent military actions, where cyber operations enabled or reduced the risk of carrying out conventional strikes, we&apos;re left wanting&#x2026; much more.&#xa0;</p><p>Take, for example, the 2025 strikes against Iranian nuclear facilities. US Cyber Command <a href="https://therecord.media/iran-nuclear-cyber-strikes-us"><u>reportedly helped blind</u></a> Iranian air defence systems. Earlier this year, the capture of Venezuelan President Nicol&#xe1;s Maduro was facilitated <a href="https://news.risky.biz/china-fights-scam-compounds-for-china/#maduro-raid-cements-disruptive-cyber-role"><u>by switching off streetlights</u></a> so that American forces could operate under the cover of darkness. And the decapitation strike against Iran&apos;s Supreme Leader Ali Khamenei was aided by an operation that disrupted nearby mobile phone towers so that his protection detail could not receive warning of an impending attack.</p><p>In each of these cases the cyber element enabled or reduced the risk of carrying out conventional components of the strikes.</p><p>By contrast, Ukrainian drone strikes have been targeting Wildberries facilities since mid-July this year and have disabled seven of the company&apos;s ten facilities <a href="https://t.me/ministry_of_defense_ua/16040"><u>according to Ukraine&apos;s defence ministry</u></a>. The impact of using cyber to disable call centres or payment systems pales in comparison to the <a href="https://news.liga.net/en/politics/news/journalists-showed-what-the-wildberries-warehouse-in-voronezh-looks-like-after-the-strike-satellite-images"><u>lasting damage of bombs</u></a>.&#xa0;&#xa0;&#xa0;</p><p>Cyber operations may not have amplified any kinetic one here, but that doesn&apos;t mean it was for nought. In fact we think the exact opposite is happening here. A successful drone strike is being used to amplify the psychological impact of a cyber attack.&#xa0;</p><p>The <a href="https://www.npr.org/2026/07/25/nx-s1-5905377/ukraine-drone-strikes-on-wildberries-warehouses-hit-russian-sellers-consumers"><u>notional justification</u></a> for Ukraine&apos;s attacks on Wildberries is that it sells a range of dual-use goods including navigation systems, components for drones and supplies for the Russian army. Perhaps more importantly, though, as with <a href="https://www.caspianpolicy.org/research/security/live-map-of-russian-refineries-hit-ukrainian-drone-strikes-boost-caspian-energy"><u>attacks on Russian oil refineries</u></a>, it helps highlight the costs of the war to ordinary Russians.</p><p>In this case, it is the psychological impact that is important. And even just saying that cyber attacks and drone strikes amplified each other is great propaganda.&#xa0;&#xa0;</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/Dy-_WEXg-IQ?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Trump&apos;s private hacker memo is the right idea"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>German bank hackers caught:</strong> German and Brazilian authorities <a href="https://www.gov.br/pf/pt-br/assuntos/noticias/2026/08/pf-deflagra-operacao-contra-fraudes-bancarias-eletronicas-e-lavagem-de-dinheiro"><u>cooperated to arrest members</u></a> of a group that allegedly stole more than &#x20ac;30 million from German online banking customers. <a href="https://www.bka.de/DE/Presse/Listenseite_Pressemitteilungen/2026/Presse2026/260814_PM_Ermittlungserfolge_Computerbetrug.html"><u>German federal police say</u></a> the group exploited a vulnerability in the systems of a payment service provider and stole the funds within just a few days in November 2023.</li><li><strong>94 scam centres shut down in Ukraine: </strong>Ukraine&apos;s<strong> </strong>cyber police <a href="https://cyberpolice.gov.ua/news/naczpolicziya-provela-ponad--obshukiv-i-prypynyla-robotu--shaxrajskyx-call-czentriv-640/"><u>disrupted the scam centres</u></a> after carrying out more than 400 searches. Authorities seized more than US&#x24;2 million and over 3,300 computers.&#xa0;</li><li><strong>Cryptocurrency exchange sues North Korea: </strong>The Bybit cryptocurrency exchange <a href="https://www.bybit.com/en/press/post/bybit-sues-north-korea-and-lazarus-group-secures-preliminary-injunction-freezing-stolen-assets-in-landmark-crypto-asset-recovery-effort-bb55bb16f1710f487aa"><u>has announced</u></a> it has filed a civil lawsuit against the North Korean government and its Reconnaissance General Bureau, the intelligence agency responsible for state-sponsored crypto hacking. North Korea stole USD&#x24;1.5 billion from Bybit last year, and the company has obtained an injunction freezing some of the stolen funds from being moved. Bybit has recovered &#x24;48 million of the stolen funds and frozen a further &#x24;30 million.&#xa0; We don&apos;t know how much joy they&apos;ll get from the suit, but we enjoy seeing novel strategies being pursued.</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Casey Ellis chats with Socket founder Feross Aboukhadijeh about npm 12&#x2019;s move to disable install scripts by default.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI140.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI140/&quot;&gt;Sponsored: What npm 12 fixes&#x2026; and what it doesn&#x2019;t&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 17:54 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: What npm 12 fixes&#x2026; and what it doesn&#x2019;t&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p></p><h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss The Offence Death Cycle, a paper looking at how to take advantage of a defender&#x2019;s ability to control a network to discover intruders.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN179.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN179/&quot;&gt;Between Two Nerds: The eye of Sauron&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 32:06 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: The eye of Sauron&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/6wxeh8ACGRY?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: The eye of Sauron"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Slovakia finds Russian backdoor in traffic speed cameras:</strong> Slovakia&apos;s national security service NBU has issued a security alert against the use of NERO R-ONE high-speed traffic cameras.</p><p>The <a href="https://www.nbu.gov.sk/varovanie-pred-rizikami-cestnych-meradiel/"><u>agency says</u></a> the cameras contain a backdoor mechanism that will execute malicious code received via an SMS from a list of hardcoded Russian phone numbers.</p><p>The NBU started an investigation into the devices after <a href="https://www.aktuality.sk/clanok/t9SbnDt/zahadne-radary-matusa-sutaja-estoka-su-ruske-alebo-z-cypru/"><u>multiple</u></a> <a href="https://dennikn.sk/5493401/nove-kamery-sutaja-estoka-su-od-cyperskej-schranky-bez-historie-a-jedina-stopa-vedie-k-rusku/"><u>reports</u></a> in Slovak media that the cameras were bought with a no-bid direct contract from a Cyprus shell company with fake certifications.</p><p>[<a href="https://news.risky.biz/risky-bulletin-slovakia-finds-russian-backdoor-in-traffic-speed-cameras/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>The EU publishes its upcoming cybersecurity standards: </strong>The European Telecommunication Standards Institute has <a href="https://www.etsi.org/newsroom/press-releases/etsi-launches-approval-process-for-17-european-standards-supporting-the-cyber-resilience-act/"><u>released</u></a> 17 cybersecurity standards that vendors will have to follow to sell products in the EU when the EU Cyber Resilience Act enters into effect in December of 2027.</p><p>The standards cover 17 core technologies for major product categories including operating systems, routers, firewalls, VPNs, browsers and password managers.</p><p>The standards describe a list of minimum security features each product category must implement to be CRA-compliant. Most of the standards usually require the same basic features, such as the ability to update the product once it&apos;s sold, that devices are sold with an SBOM, that they use modern cryptography, or that they ship with secure-by-default settings.</p><p>[<a href="https://news.risky.biz/risky-bulletin-the-eu-publishes-its-upcoming-cybersecurity-standards/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>AI agents had turf wars:</strong> Anthropic says agents &quot;consistently&quot; engaged in a turf war when they received the same tasks. Agents deployed self-replicating malware, locked rival accounts, and deployed scripts that ran in a loop to disable a competing agent&apos;s processes. Anthropic says the newer models like Mythos won by revoking rivals&apos; access first and then negotiating truces. [<a href="https://www.anthropic.com/research/multiagent-systems"><u>Anthropic</u></a>]</p>
  124. Srsly Risky Biz: Data Theft Extortion Is Booming! Hooray!

    Thu, 13 Aug 2026 09:46:41 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.island.io/"><em><u>Island</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58-1.jpeg" alt="Srsly Risky Biz: Data Theft Extortion Is Booming! Hooray!"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.island.io/"><em><u>Island</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB179.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB179/&quot;&gt;Srsly Risky Biz: Data extortion is booming. Hooray!&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 30:48 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Data extortion is booming. Hooray!&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58.jpeg" class="kg-image" alt="Srsly Risky Biz: Data Theft Extortion Is Booming! Hooray!" loading="lazy" width="2000" height="1359" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-50338bd0-0a35-4449-a412-c6d956978b58.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@jensholm?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">oskar holm</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/a-bird-flying-in-the-sky-e7ZuyidGtKQ?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>The cybercriminal ecosystem is increasingly focussing on data theft and extortion rather than locking up victims&apos; files. That criminals have stumbled across a new lucrative business model is a bittersweet win in the fight against disruptive, encrypting ransomware. Data theft extortion isn&apos;t great, but it doesn&apos;t leave widespread chaos in its wake like ransomware can.</p><p>Silent Ransom, aka Luna Moth, is one group currently making big bucks from data theft extortion. Last week The <em>Cyber Risk Insurer</em> <a href="https://www.theinsurer.com/cyber-risk/news/exclusive-goodwin-procter-paid-around-10-million-ransom-to-luna-moth-with-brit-2026-08-07/"><u>reported two law firms</u></a> had <a href="https://www.theinsurer.com/cyber-risk/news/exclusive-wilmerhale-paid-at-least-18-million-ransom-to-luna-moth-as-cna-covered-2026-08-07/"><u>paid substantial ransoms</u></a> to the group this year: Goodwin Procter and WilmerHale, which paid USD&#x24;10 million and &#x24;USD18 million respectively.&#xa0;</p><p>Silent Ransom has been targeting law firms <a href="https://www.ic3.gov/CSA/2025/250523.pdf"><u>since 2023</u></a>. It historically used phishing and convinced victims to install legitimate remote access software which was then used to steal sensitive data. In the last year, however, they&apos;ve brazenly <a href="https://www.ic3.gov/CSA/2026/260526.pdf"><u>sent people</u></a> to <a href="https://edition.cnn.com/2026/06/27/politics/cybercriminals-hire-burglars-russian-us-law-firms"><u>compromise systems in person</u></a> by posing as IT support staff.</p><p>Its data exfiltration process prioritises speed over completeness. Google&apos;s Threat Intelligence Group (GTIG) <a href="https://cloud.google.com/blog/topics/threat-intelligence/targeted-campaign-us-law-firms"><u>says the groups&apos; entire attack process</u></a>, from initial target contact to data theft and extortion, is often completed within a single day. With ransom payments in the millions, that works out to a pretty good hourly rate.</p><p>Another group that <a href="https://cloud.google.com/blog/topics/threat-intelligence/blackfile-vishing-extortion-operation/"><u>emerged in early 2026</u></a> was BlackFile, which now calls itself Redact. The group has also honed in on targeting similar organisations, but its data exfiltration is more comprehensive.</p><p>Redact gains initial access using sophisticated high-volume voice phishing attacks (vishing) to steal credentials from victim organisations. It then uses these credentials to steal data from OneDrive and Sharepoint. It also pivots out to other software-as-a-service applications.&#xa0;</p><p><a href="https://cloud.google.com/blog/topics/threat-intelligence/unc6671-targets-financial-services-and-enterprise-cloud-environments"><u>GTIG says</u></a> BlackFile spent April and May this year targeting enterprises in the real estate, healthcare and insurance sectors. In June it had shifted its focus to large technology, transportation and hospitality organisations, particularly those with valuable intellectual property, source code or sensitive VIP client data. By July it had zeroed in on the financial and legal sectors including private equity firms, law firms and financial rating agencies.&#xa0;</p><p>Just last week <em>Reuters</em> reported a <a href="https://www.reuters.com/legal/government/major-wall-street-hedge-funds-targeted-attempted-cyberattacks-bloomberg-news-2026-08-05/"><u>series of attempted cyber attacks</u></a> against Wall Street. These attacks, linked <a href="https://www.bleepingcomputer.com/news/security/hedge-fund-cyberattacks-tied-to-blackfile-linked-unc6671-extortion-group/"><u>to BlackFile</u></a>, targeted several major hedge funds and private equity firms.&#xa0;</p><p>Based on Bitcoin transactions Google estimates BlackFile collected more than USD&#x24;10 million in payments from February to mid-May. The average ransom was USD&#x24;750,000.&#xa0;</p><p>The cybercrime ecosystem really does seem to be migrating from encrypting ransomware to data theft extortion. Exact numbers vary, but <a href="https://unit42.paloaltonetworks.com/cyber-extortion-economy/"><u>a number</u></a> of <a href="https://cyberscoop.com/google-threat-intelligence-group-ransomware-report-2026/"><u>cyber security</u></a> and <a href="https://4526209.fs1.hubspotusercontent-na1.net/hubfs/4526209/2025%20Claims%20Report.pdf"><u>risk firms</u></a> have found data theft extortion rates are up while encrypting ransomware <a href="https://www.sophos.com/en-us/content/state-of-ransomware"><u>rates are down</u></a>.&#xa0;</p><p>This is good news and there are a number of potential reasons for this shift.</p><p>Victim organisations have <a href="https://www.sans.org/blog/stay-ahead-ransomware-evolution-from-encryption-to-extortion"><u>gotten better at recovering from backups</u></a>, for a start. And for cyber criminals, stealing data is less work than building and supporting encrypting malware. The threat of publishing stolen data is a very powerful motivator in certain verticals, too, and that threat can&apos;t be mitigated by maintaining good backups.&#xa0;</p><p>Law firms, for example, are reputation-based businesses. The threat of confidential and attorney-client privileged information being leaked could easily be an existential threat to a firm. As Google <a href="https://cloud.google.com/blog/topics/threat-intelligence/targeted-campaign-us-law-firms"><u>puts it</u></a>, law firms &quot;may be highly motivated to resolve extortion situations quietly to protect their professional standing&quot;. They&apos;re not wrong!</p><p>We described this operational shift to data theft extortion as a win at the beginning of this article because it doesn&apos;t have the disruptive impact that ransomware often can.&#xa0;</p><p>That&apos;s not to say that encrypting ransomware isn&apos;t still a problem. It hasn&apos;t gone away completely and we&apos;re still seeing some massive incidents from time to time. Last year&apos;s <a href="https://news.risky.biz/srsly-risky-biz-america-wont-beat-the-distillation-ecosystem/#:~:text=Jaguar%20Land%20Rover%20Hackers%20Were"><u>Jaguar Land Rover incident</u></a> had a measurable impact on the UK economy and the 2024 <a href="https://www.kaspersky.com.au/blog/unitedhealth-ransomware-attack/34673/"><u>UnitedHealth ransomware incident</u></a> had <a href="https://en.wikipedia.org/wiki/Change_Healthcare#2024_cyberattack"><u>serious effects</u></a> on US healthcare services.&#xa0;</p><p>With this in mind, governments should double down on countering encrypting ransomware to further incentivise criminals towards these lower-impact extortion campaigns. Disruption operations targeting the encryption ransomware gangs causing catastrophic results should be maintained. But the data extortion groups? Let them eat their multi-million dollar cake. At least for now.</p><p>Data theft extortion is the new hotness, and it may even be sucking the air out of encrypting ransomware. That&apos;s even more reason for governments to maintain the pressure on the ransomware ecosystem. The balloon is deflating. It&apos;s time to squeeze even harder.&#xa0;</p><h2 id="secure-by-design-deserves-a-resurrection">Secure by Design Deserves a Resurrection</h2><p>Advances in AI models and a maturing understanding of how to harness them to write secure code mean it&apos;s the perfect time to reinvigorate CISA&apos;s stalled <a href="https://www.cisa.gov/securebydesign"><u>Secure by Design initiative</u></a>.</p><p>The idea behind Secure by Design was simply to encourage companies to develop more secure products using carrots and sticks. The effort involved creating secure software development standards for companies to follow. The plan was to eventually introduce legislation that made companies liable if they continued to churn out insecure crapware. Companies that adhered to the standards would be fine, and those that didn&apos;t would suffer.</p><p><a href="https://news.risky.biz/grandpa-biden-cyber-president-65c0eb42a0e6da001a37dfd9/"><u>When we first wrote about Secure by Design</u></a> three years ago we liked the idea but thought it would be a very long-term effort that would take at least a decade to implement. Secure software development was slow and arduous and even with clear guidance a lot of industry was always going to push back on the initiative.</p><p>President Donald Trump&apos;s Cyber Strategy <a href="https://www.whitehouse.gov/wp-content/uploads/2026/03/President-Trumps-Cyber-Strategy-for-America.pdf"><u>was not keen</u></a> on strengthening regulation and although the Secure by Design initiative wasn&apos;t killed outright, it has certainly <a href="https://www.cybersecuritydive.com/news/cisa-secure-by-design-initiative-limbo-key-leaders-resign/745986/"><u>withered on the vine</u></a>.&#xa0;</p><p>Things have since changed. The rise of very capable AI now means it is possible to take advantage of AI to produce more secure code from the get-go. It&apos;s <a href="https://www.microsoft.com/en-us/microsoft-copilot/for-individuals/do-more-with-ai/general-ai/what-are-ai-tokens"><u>just tokens</u></a>, not real-world blood, sweat and tears.</p><p>Late last month Google <a href="https://blog.google/security/chrome-stronger-with-every-update/"><u>described how</u></a> it was using AI models in its development practices to both discover existing bugs in Chrome and prevent new ones from being introduced. This isn&apos;t as simple as telling Gemini to run off and write secure code with no mistakes. Google uses harnesses to run separate multi-agent workflows that find vulnerabilities, triage them, and then fix them. And in May this year Microsoft rolled out what it calls MDASH, its multi-model agentic vulnerability discovery harness.&#xa0;</p><p>So the tech giants are already using AI to level-up their code quality, both through bug discovery and, perhaps more importantly, prevention.</p><p>We also think it might be time to revisit the original Secure by Design plan to make vendors at least somewhat liable for flaws in their products, too. The current status quo means vendors can absolve themselves of almost all liability via End User License Agreements (EULAs), a situation that probably made sense when expertise to discover vulnerabilities was rare.&#xa0;</p><p>Now, however, it looks like the smart use of AI can be used to eliminate entire classes of security vulnerabilities. Why should software vendors remain completely unaccountable if improving the security of their products is more straightforward?</p><p><a href="https://www.csis.org/analysis/end-could-ai-undermine-softwares-us-legal-shield"><u>This Center for Strategic and International Studies report</u></a> released last week contains useful suggestions for lawmakers when it comes to reshaping software liability. These include limiting liability for software producers that follow recognised secure development practices and drawing a distinction between open source maintainers and the for-profit companies that take advantage of their software.&#xa0;</p><p>AI is changing software development, making it faster and cheaper. But it needs a little government push to also make it more secure.</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/pI1m-7letn4?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Data extortion is booming. Hooray!"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Iranian attacks galvanise water security initiatives:</strong> Initiatives from lawmakers <a href="https://fedscoop.com/treasury-tax-code-ot-cyberattacks-tom-cotton-letter/"><u>include suggesting</u></a> that the tax code be amended to encourage investment in operational technology to better secure American critical infrastructure. A proposed <a href="https://www.schiff.senate.gov/news/press-releases/news-sens-schiff-klobuchar-release-comprehensive-cybersecurity-bill-in-wake-of-recent-cyberattacks-on-water-infrastructure/"><u>Water Cyber Shield Act</u></a> would also provide more money and authorise the Environmental Protection Agency to carry out cyber security assessments. The volunteer DEF CON Franklin project <a href="https://www.cybersecuritydive.com/news/water-cybersecurity-mdr-services-def-con-franklin/827449/"><u>also announced</u></a> that it will pay for cyber security firms to provide monitoring and protection for small water utilities. It has some initial seed funding but is seeking more grants and ultimately thinks it will need federal government money to scale up operations.&#xa0;</li><li><strong>FBI collaborate with China, Russia:</strong> <em>Reuters</em> <a href="https://www.reuters.com/world/china/under-patel-fbi-forges-unprecedented-law-enforcement-ties-with-china-russia-2026-08-05/"><u>reports</u></a> that the FBI is collaborating with Chinese and Russian authorities to combat specific transnational crimes such as fentanyl trafficking, cyber scams and child sexual exploitation. These are serious problems that all countries involved have an interest in tackling. Hopefully collaboration will end up with some wins. At the same time, Russia and China are America&apos;s adversaries, so there are real counterintelligence concerns. Not to worry, we are sure FBI Director Kash Patel will mitigate these risks in a <a href="https://www.thedailybeast.com/keystone-kash-patels-endless-wild-booze-binges-spilled-by-top-officials/"><u>serious and sober</u></a> way.&#xa0;</li><li><strong>Supply chain protection for open source maintainers:</strong> Open source supply chain security company Socket is <a href="https://socket.dev/blog/free-business-plan-upgrades-for-open-source?utm_medium=feed"><u>offering free business plans</u></a> to open source maintainers. <em>Disclaimer: Socket is a Risky Business sponsor.</em>&#xa0;&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Catalin Cimpanu talks with Michael Leland, Field CTO of Island, about the company&apos;s seamless expansion into SASE and enterprise AI.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI139.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI139/&quot;&gt;Sponsored: Island&apos;s expansion to SASE and enterprise AI&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 16:59 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Island&apos;s expansion to SASE and enterprise AI&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about examples of cyber resistance and whether they achieve their goals.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN178.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN178/&quot;&gt;Between Two Nerds: The cyber resistance!&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 29:53 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: The cyber resistance!&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/eRJWUKLQK6M?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: The cyber resistance!"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Russian hackers adopt the fake job interview tactics:</strong> One of Russia&apos;s elite military hacker groups is targeting system administrators and IT professionals in Ukraine using fake job interviews as a malware delivery vector.</p><p>Ukraine&apos;s CERT says the campaign began in May and is ongoing.</p><p>The attacks have been linked to UAC-0145, a sub-group of Sandworm, a veteran cyber unit inside Russia&apos;s GRU military intelligence agency.</p><p>Using fake job interviews as a malware delivery vector has been a popular tactic used by Iran and North Korea for several years. It&apos;s been used initially for cyber-espionage, but is now a popular entry vector for many of North Korea&apos;s financially-motivated operations, especially the ones targeting the cryptocurrency and Web3 developer community. China has <a href="https://www.mi5.gov.uk/five-eyes-joint-bulletin-safeguarding-our-secrets"><u>also done this</u></a>, but they usually use job interviews to recruit insiders, rather than hack their targets.</p><p>[<a href="https://news.risky.biz/risky-bulletin-russian-hackers-adopt-the-fake-job-interview-tactics/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Pwnie Awards 2026 winners:</strong> If there&apos;s one thing that has annoyed me as someone who doesn&apos;t attend the yearly BlackHat &amp; DEFCON conferences, it&apos;s that I could never find out who won the Pwnie Awards for days and sometimes weeks after the event had concluded.But not this year, thanks to TechCrunch&apos;s Zack Whittaker, who <a href="https://this.weekinsecurity.com/microsoft-wins-lamest-vendor-at-pwnie-awards-2026-for-threatening-security-researchers-with-legal-action/"><u>spotted and shared</u></a> a <a href="https://www.youtube.com/live/8Eo4ZDQTAFs?t=4685s"><u>private link</u></a> to one of the DEFCON live streams that broadcast the Pwnie Awards ceremony.</p><p>Let&apos;s just say curiosity got the better of me and I spent an hour extracting and sharing a list of this year&apos;s winners.</p><p>[<a href="https://news.risky.biz/risky-bulletin-pwnie-awards-2026-winners/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Meta&apos;s AI joins Anthropic and OpenAI in the hacky-hacky:</strong> The UK&apos;s AI Security Institute has disclosed a security incident after two AI models it was evaluating performed actions the agency wasn&apos;t expecting and tried to hack real-world organizations.</p><p>The incident took place at the end of last month and the malicious actions were performed by Anthropic&apos;s Mythos 5 and OpenAI&apos;s GPT-5.6-Sol models, which were among the several models being tested at the time.</p><p>[<a href="https://news.risky.biz/risky-bulletin-metas-ai-joins-anthropic-and-openai-in-the-hacky-hacky/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  125. Srsly Risky Biz: Being a North Korean Hacker Is About to Be Less Fun

    Thu, 06 Aug 2026 04:33:06 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://permiso.io/"><em><u>Permiso Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpg" alt="Srsly Risky Biz: Being a North Korean Hacker Is About to Be Less Fun"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://permiso.io/"><em><u>Permiso Security</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB178.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB178/&quot;&gt;Srsly Risky Biz: Being a North Korean hacker is about to be less fun&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 22:56 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Being a North Korean hacker is about to be less fun&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="srsly-risky-biz-being-a-north-korean-hacker-is-about-to-be-less-fun">Srsly Risky Biz: Being a North Korean Hacker Is About to Be Less Fun</h2><figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpeg" class="kg-image" alt="Srsly Risky Biz: Being a North Korean Hacker Is About to Be Less Fun" loading="lazy" width="2000" height="1454" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/08/data-src-image-d72bc376-ac72-44ec-b754-9bb04d980ae5.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><i><em class="italic" style="white-space: pre-wrap;">Photo by</em></i><a href="https://unsplash.com/@amaitu?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><i><em class="italic underline" style="white-space: pre-wrap;">Steve Barker</em></i></u></a><i><em class="italic" style="white-space: pre-wrap;"> on</em></i><a href="https://unsplash.com/photos/white-concrete-building-under-blue-sky-during-daytime-z89pcnApqfk?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><i><em class="italic underline" style="white-space: pre-wrap;">Unsplash</em></i></u></a></figcaption></figure><p>North Korea will have to rein in its pet hackers after seemingly losing control over them.&#xa0;</p><p>Last week we <a href="https://news.risky.biz/srsly-risky-biz-chipping-away-at-chinese-ai-risks/#north-korean-hackers-hack-north-korea"><u>covered the news</u></a> that a group of former North Korean military intelligence operatives had been caught hacking into the country&apos;s banks to steal funds for their personal benefit.&#xa0;</p><p><em>Daily NK</em> <a href="https://www.dailynk.com/english/north-korea-elite-bank-hacking-ring-arrested/"><u>reports</u></a> Pyongyang&apos;s elite are shocked at &quot;the scale and audacity of the scheme&quot;. Punishment for those involved will reportedly be extreme, with one official saying &quot;It will be hard for the entire family line to survive&quot;. Grim.</p><p>The outlet also says officials in North Korea&apos;s <a href="https://en.wikipedia.org/wiki/Reconnaissance_General_Bureau"><u>Reconnaissance General Bureau</u></a>, the organisation responsible for cyber espionage, are worried that the scandal could even roll uphill and claim <em>their</em> scalps.</p><p>We expect, therefore, that this incident will result in some substantial tightening of operational controls.</p><p>North Korea&apos;s state-sanctioned cyber crime operations can be lumped into three buckets: high-value targeted <a href="https://www.chainalysis.com/blog/crypto-hacking-stolen-funds-2026/"><u>cryptocurrency hacks</u></a>, broad-based <a href="https://www.dfat.gov.au/news/media-release/joint-statement-dprk-it-workers"><u>fraudulent worker scheme</u></a>, and ransomware extortion operations.</p><p>There is some evidence that in addition to former operatives stealing from North Korean banks, the government there is also losing control over its ransomware operators.</p><p>North Korea first dipped its toe into ransomware by developing its own strains. In 2021 and 2022 one of its hacker groups, Andariel, <a href="https://cloud.google.com/blog/topics/threat-intelligence/apt45-north-korea-digital-military-machine"><u>created two different ransomware strains</u></a> that it <a href="https://securelist.com/andariel-evolves-to-target-south-korea-with-ransomware/102811/"><u>used on organisations</u></a> in the US, South Korea and Japan, including <a href="https://www.cisa.gov/news-events/cybersecurity-advisories/aa22-187a"><u>against the American health sector</u></a>.&#xa0;</p><p>After dabbling in roll-your-own ransomware, in recent years North Korean hackers began collaborating with the ransomware-as-a-service (RaaS) ecosystem. Threat intelligence reports say that Andariel <a href="https://unit42.paloaltonetworks.com/north-korean-threat-group-play-ransomware/"><u>used Play ransomware</u></a> in 2024 and <a href="https://www.security.com/threat-intelligence/lazarus-medusa-ransomware"><u>Medusa ransomware in 2025</u></a>.&#xa0;</p><p>Last week, South Korean security firm AnhLab <a href="https://image.ahnlab.com/atip/content/file/20260730/%5BAhnLab%5DOperation%20Double%20Barrel(ENG)(2026.07.30).pdf"><u>reported that</u></a> a state-controlled North Korean hacker group is also collaborating with <a href="https://www.cloudsek.com/blog/inside-gunra-raas-from-affiliate-recruitment-on-the-dark-web-to-full-technical-dissection-of-their-locker"><u>Gunra ransomware</u></a>. It&apos;s unclear whether this collaboration is state sanctioned or not.</p><p>North Korean cyber units are supposed to funnel stolen funds into the state&apos;s coffers. By design, however, RaaS offerings help skilled individuals profit from their hacks.&#xa0;</p><p>Given the recent hack of the country&#x2019;s banks, and this news of DPRK operators getting cozy with criminal outfits, we suspect that RGB officials might start to look at RaaS operations a bit differently. If you&apos;re personally on the hook for extreme punishments if your underlings go rogue, why put them in a position where they&apos;re tempted to put their hands in the cookie jar?</p><p>So what&apos;s next for North Korean state-sponsored hacks? Scaling back its ransomware operations is one possibility, but it is not the only option. We suspect a strong tightening of controls on its hacking teams in general is more likely.</p><p>There is good evidence that North Korea uses very tight internal controls on its scam IT workers already. North Korean defectors <a href="https://www.tovima.com/wsj/a-defector-explains-the-remote-work-scam-helping-north-korea-pay-for-nukes/"><u>have described</u></a> constant surveillance and screen monitoring, isolation, movement restrictions and strict work quotas.&#xa0;</p><p>It doesn&apos;t appear that the same level of control is currently applied to the country&apos;s hackers. A <a href="https://www.nbcnews.com/tech/security/north-korea-hackers-are-handpicked-pampered-elite-reuters-n262396"><u>2014 report</u></a> says they were viewed as the elite of the military. Rather than being intensively surveilled, they had privilege and more freedom.&#xa0;</p><p>That&apos;s almost certainly a thing of the past. Being a state-backed DPRK hacker is about to be a lot less fun.</p><h2 id="cyber-war-is-here-and-america-is-politically-unprepared">Cyber War Is Here, and America Is Politically Unprepared</h2><p>Multiple cyber provocations targeting America&apos;s water sector have revealed how unprepared the US is to deal with the political fallout of cyber attacks against critical infrastructure.</p><p><em>The Washington Post</em> <a href="https://www.washingtonpost.com/national-security/2026/07/30/us-spy-agencies-suspect-iran-launched-cyberattack-minnesota-water-facilities/"><u>reported</u></a> US intelligence agencies have determined that Iran is behind the attacks on water facilities in Minnesota that we <a href="https://news.risky.biz/srsly-risky-biz-chipping-away-at-chinese-ai-risks/#irans-perfectly-calibrated-cyber-attack"><u>covered last week</u></a>. Now, <a href="https://abcnews.com/US/12-states-face-cyberattacks-water-systems-sources/story?id=135348482"><u>at least a dozen states</u></a> have experienced similar disruptions. <a href="https://news.risky.biz/risky-bulletin-hacker-breaches-hungarys-state-treasury/"><u>Per <em>Risky Bulletin</em></u></a>:</p><blockquote>Some of the new public incidents have been reported in Clayton County, Georgia and the city of Duchesne, Utah. Customers were left low pressure or no water in Clayton County in the middle of the night last week. In Utah, the cyberattack made pumps run dry while their control panels said they were pumping water. The incident impacted an oilfield wastewater disposal site but did not cause any environmental damage.&#xa0;</blockquote><p>A CISA advisory <a href="https://www.cisa.gov/news-events/alerts/2026/07/30/cisa-urges-water-and-wastewater-systems-sector-protect-ot-against-activity-targeting-plcs"><u>about the escalating activity</u></a> says that it has &quot;resulted in boil water notices&quot; and led to &quot;sustained manual operations&quot;.</p><p>This campaign is historically significant. As far as we know, it is the first time a country has responded to kinetic attacks in the cyber domain by targeting an aggressor&apos;s critical infrastructure.&#xa0;</p><p>The direct impact of the hacks on water supply have been manageable so far, which is consistent with what we&apos;ve seen in other conflicts over the last several years.<em>&#xa0;</em></p><p>The effects of wartime cyber attacks are relatively short-lived and they have been most useful when combined with tightly orchestrated conventional operations, such as America&apos;s 2025 strikes <a href="https://news.risky.biz/srsly-risky-biz-microsofts-forgoes-its-secure-future/#%5C34%2004-iranian-air-defence-not-found"><u>against Iranian nuclear facilities</u></a> or its <a href="https://news.risky.biz/china-fights-scam-compounds-for-china/#maduro-raid-cements-disruptive-cyber-role"><u>capture of</u></a> Venezuelan President Nicolas Maduro. Without complementary conventional action, Iran&apos;s attacks on America&apos;s water systems don&apos;t amount to much.&#xa0;</p><p>Having said that, the strategic goal of the campaign is to erode political support for the war. In our view, widespread publicity without accompanying devastation is the perfect way to achieve that goal.</p><p>This campaign was entirely predictable. As <a href="https://news.risky.biz/srsly-risky-biz-successful-war-leaves-iran-with-one-option-cyber/"><u>we&apos;ve written before,</u></a> if you bomb Iran, you should expect cyber operations against critical infrastructure in return. Even as far back as 2013 Iranian hackers compromised control systems at a New York state dam and tried to affect its operation.&#xa0;</p><p>We&apos;re only surprised that this current campaign took so long to spin up.&#xa0;</p><p>The US government&apos;s response has so far been fairly muted. CISA has advised organisations to remove targeted devices from the internet and to connect to them using VPNs or gateway devices, to enable passwords and use strong ones, and allowlist IP addresses for remote access.</p><p>Since it was pretty clear that Iran would respond to missiles with cyber, we think running a campaign to fix these vulnerabilities <em>before</em> military action against Iran would have left the US government in a much better place to deal with this campaign.</p><p>It&apos;s true that a preparatory cyber security drive like this wouldn&apos;t have made America&apos;s water infrastructure perfectly secure. It&apos;s too big, too decentralised and too resource constrained. It could, however, have made a <em>political</em> difference. &quot;We understand the problem, we&apos;ve been putting mitigations in place, some systems remain vulnerable but we can assure everyone the impacts will be limited&quot;. In other words: We&apos;ve got this under control.</p><p>As it stands, despite the campaign&apos;s predictability, the Trump administration is on the back foot. President Donald Trump even <a href="https://cyberscoop.com/trump-blames-minnesota-water-cyberattacks-iran/"><u>blamed Minnesota</u></a> and its officials for the attacks.&#xa0;</p><p>That&apos;s some pretty bonkers politics, and exactly the sort of sound byte Iran will chalk up in the &quot;win&quot; column.</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/t3qbfsllBKA?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Being a North Korean hacker is about to be less fun"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Chrome in the AI era:</strong> Google&apos;s Chrome team <a href="https://blog.google/security/chrome-stronger-with-every-update/"><u>has described</u></a> how they are using AI to make the browser safer by improving vulnerability discovery and patching. The team uses AI agents coupled with harnesses to carry out find and fix vulnerabilities as well as to triage external bug reports. Google says that over the previous two stable releases it fixed 1072 security bugs.&#xa0;</li><li><strong>OpenAI disrupts Cambodian scammers:</strong> OpenAI <a href="https://openai.com/index/disrupting-malicious-uses-of-ai-criminal-scam-operation/"><u>announced last week</u></a> that it had disrupted a Cambodia-based scam operation that was using ChatGPT to support &quot;investment, romance, gambling and impersonation schemes&quot;. The good news here is that a tip from WhatsApp led OpenAI to what it found to be a coordinated network of accounts. We are all in favour of more coordination to counter scam compounds.</li><li><strong>Romance scammer gets seven years:</strong> The US Department of Justice <a href="https://www.justice.gov/usao-sdny/pr/ghanaian-national-sentenced-85-months-prison-stealing-more-10-million-romance-scams"><u>announced</u></a> last week that Derrick Van Yeboah was sentenced to 85 months in prison for his role in romance and business email compromise scams. The Ghanaian was involved in a criminal organisation that stole and laundered more than USD&#x24;100 million from dozens of victims. The DoJ says Van Yeboah &quot;personally perpetrated many of the romance scams by impersonating fake romantic partners&quot; with victims. His victims transferred millions of dollars to the gang.&#xa0;&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Permiso CTO Ian Ahl about detecting ShinyHunters-style attackers as they move through cloud and SaaS environments.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI138.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI138/&quot;&gt;Sponsored: The intrusion signals hiding in plain sight&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 15:36 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: The intrusion signals hiding in plain sight&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="shorts">Shorts</h2><h3 id="frontier-lab-cybersecurity-testing-controls-are-rubbish">Frontier Lab Cybersecurity Testing Controls Are Rubbish</h3><p>In recent weeks <a href="https://openai.com/index/hugging-face-model-evaluation-security-incident/"><u>OpenAI</u></a> and <a href="https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals"><u>then Anthropic</u></a> announced that their AI agents have, ahem, exceeded expectations in different testing scenarios. In both cases, AI models escaped notional testing sandboxes, and the companies detected the escapes days to weeks after the fact.&#xa0;&#xa0;</p><p>By contrast, the UK&apos;s AI Security Institute (AISI) <a href="https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing"><u>wrote this week</u></a>:&#xa0;</p><p>On 28th July 2026, AISI&apos;s Security Team detected unusual data transfers leaving our research systems during a routine cyber evaluation. On investigation, we found that some of the agents being tested had engaged in sustained, potentially harmful activity directed at real people and organisations. We declared a security incident and, <strong>within roughly one hour of discovery</strong>, had contained it and begun a full investigation. [emphasis added]</p><p>How refreshingly competent!</p><p>AISI&apos;s write up covers essentially the same type of behaviour as described by OpenAI and Anthropic, but it is far better. It states things plainly without guff and corpo-speak.&#xa0;</p><h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> talk about whether hacker culture is inherently anti-authoritarian and how different states get their country&apos;s hackers to work for them.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN177.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN177/&quot;&gt;Between Two Nerds: Hackers vs the state&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 25:50 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Hackers vs the state&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/7vYo2OZIAZA?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Hackers vs the state"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Hacker breaches Hungary&apos;s State Treasury: </strong>The same hacker who hit and wiped Romania&apos;s land registry database has now hacked Hungary&apos;s State Treasury in another brazen intrusion into an extremely sensitive government system.</p><p>The incident took place last week and portions of the stolen data have since been put up for sale on an underground hacking forum.</p><p>The intrusion was <a href="https://telex.hu/techtud/2026/08/02/magyar-allamkincstar-nemzeti-kifizeto-ugynokseg-kibertamadas-orosz-szerver-titkositott-allomanyok"><u>confirmed</u></a> to local journalists by Hungary&apos;s State Treasury over the weekend.&#xa0;</p><p>[<a href="https://news.risky.biz/risky-bulletin-hacker-breaches-hungarys-state-treasury/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Russia is behind the recent hotel WiFi hacks:</strong> A Russian state-sponsored hacking group is behind a recent hacking wave that has targeted and compromised hotel WiFi gateways across the globe.</p><p><a href="https://www.microsoft.com/en-us/security/blog/2026/07/31/captivecrunch-midnight-blizzard-targets-travelers-worldwide-for-malware-delivery-and-credential-theft/"><u>Microsoft says</u></a> the campaign is far larger and more complex than it was initially covered in a <a href="https://reliaquest.com/blog/threat-spotlight-dns-poisoning-tactics-expand-to-hospitality/"><u>ReliaQuest report</u></a> two weeks ago.</p><p>ReliaQuest said the hackers were modifying DNS traffic on hotel networks to redirect users to Microsoft-themed phishing sites. Microsoft says the attacks also redirected users to malware downloads, often using ClickFix pages to trick users into downloading and running the payloads.</p><p>[<a href="https://news.risky.biz/risky-bulletin-russia-is-behind-the-recent-hotel-wifi-hacks/"><u>more</u></a> on Risky Bulletin]</p><p><strong>Non-profit offers &#x24;22,000 bounty for INC ransomware group: </strong>An international crime-fighting non-profit organization is offering a <a href="https://www.crimestoppersinternational.org/bounty"><u>&#x24;22,000 bounty</u></a> for any information on members of the INC ransomware group.</p><p>To be eligible for a payout, the provided information must lead to the identification, arrest, or disruption of the gang&apos;s operations.</p><p>Crime Stoppers International is the international branch of <a href="https://en.wikipedia.org/wiki/Crime_Stoppers"><u>Crime Stoppers</u></a>, a US foundation that was established in the 70s to allow anonymous and private individuals to provide aid in US law enforcement investigations that may lack manpower or resources.</p><p>[<a href="https://news.risky.biz/risky-bulletin-non-profit-offers-22-000-bounty-for-inc-ransomware-group/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  126. Srsly Risky Biz: Chipping Away at Chinese AI Risks

    Thu, 30 Jul 2026 07:12:10 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.airlockdigital.com/"><em><u>Airlock Digital</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505-1.jpeg" alt="Srsly Risky Biz: Chipping Away at Chinese AI Risks"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://www.airlockdigital.com/"><em><u>Airlock Digital</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB177.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB177/&quot;&gt;Srsly Risky Biz: Chipping away at Chinese AI risks&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 23:18 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Chipping away at Chinese AI risks&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505.jpeg" class="kg-image" alt="Srsly Risky Biz: Chipping Away at Chinese AI Risks" loading="lazy" width="2000" height="1333" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c75f1b1d-85c5-4d21-af2c-23ab2e968505.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@ly0ns?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Li Yang</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/architectural-photograph-of-lighted-city-sky-5h_dMuX_7RE?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>The Trump administration has been trying to address two separate AI-related risks in recent months: The <em>specific</em> risk to US national security from China developing powerful AI and the <em>global</em> risk that powerful hacking machines will be available to all and sundry. A proposed bill suggests a sensible way for the US to chip away at both these risks, at least a little.</p><p>The Collaboration on Adversarial Threats and Security Risks Act <a href="https://www.banks.senate.gov/news/press-releases/sens-banks-and-schiff-introduce-bill-to-help-american-ai-companies-combat-chinese-espionage/"><u>proposes safe harbor provisions</u></a> for AI companies so they can share information related to AI-specific security risks. The idea here is to encourage frontier labs to work together to counter threats from Chinese AI labs, particularly what they describe as IP theft via <a href="https://news.risky.biz/srsly-risky-biz-us-vows-to-fight-distillation-attacks/"><u>distillation</u></a>. Without this bill, sharing this kind of information could fall afoul of anti-trust legislation that prohibits collusion.&#xa0;</p><p>We know the frontier labs can already detect distillation because they&apos;re always complaining about it after the fact. The idea behind this bill is that more permissive information sharing would let them respond quicker and disrupt at least some distillation campaigns.</p><p>A bill like this isn&apos;t necessarily the <em>most</em> effective way for the US government to prevent or deter distillation attacks. But the more direct approaches, like adding Chinese companies to the entities list, come with problems.&#xa0;</p><p>American companies, for example, are <a href="https://techcrunch.com/2026/03/22/cursor-admits-its-new-coding-model-was-built-on-top-of-moonshot-ais-kimi/"><u>increasingly</u></a> using <a href="https://opensource.org/ai/open-weights"><u>open-weight</u></a> AI models, the best of which are produced by the same Chinese labs that are carrying out large scale distillation campaigns.&#xa0;</p><p>Indeed, after Treasury Secretary Scott Bessent issued a statement saying that &quot;Entity List designations will be on the table&quot;, a tech industry coalition of 76 companies published <a href="https://www.microsoft.com/en-us/corporate-responsibility/topics/open-weight/"><u>an open letter</u></a> supporting the role of open-weight AI models in the technology mix.&#xa0;</p><p>So, the consensus seems to be that punitive action targeting Chinese labs could be a bonus for the American frontier AI companies, but likely a negative for the US economy writ large.</p><p>Really, though, the battle for AI dominance is being fought around hardware access, not the models themselves. The US controls chip exports, and there is <a href="https://www.semafor.com/article/07/28/2026/chinese-tech-grapples-with-compute-shortage"><u>some evidence</u></a> that these controls are having an impact.&#xa0;</p><p>When AI-focussed journalists Lily Ottinger and Kai Williams <a href="https://linguasinica.substack.com/p/notes-from-a-trip-to-chinas-ai-labs"><u>toured Chinese AI</u></a> labs in May this year <a href="https://blog.readsail.com/p/we-spent-10-days-touring-chinese"><u>they reported</u></a>:</p><blockquote>Basically every AI researcher we talked to in China brought up the same complaint: their companies lack sufficient access to the advanced compute resources necessary to train and run AI models.</blockquote><p>American chip export controls <a href="https://www.politico.com/news/2026/07/08/trump-ai-tech-chips-exports-00989167"><u>are imperfect</u></a>, but they&apos;re biting, and reinforcing them is the most effective way for the US to stay ahead.&#xa0;</p><p>In the areas around AI that <em>aren&apos;t</em> about staying ahead, we think there&apos;s room for the United States and China to cooperate.&#xa0;</p><p>When open-weight models get really good, we suspect neither government will be happy with everyone on the planet having access to powerful AI hacking machines. That time could only be a matter of months away, as the recently released open-weight Chinese Kimi K3 model is roughly equivalent to US frontier models from about six months ago, according to a <a href="https://www.nist.gov/news-events/news/2026/07/uk-aisi-caisi-preliminary-assessment-kimi-k3s-cyber-capabilities"><u>joint US/UK evaluation</u></a>.</p><p>China already has strict AI regulation, but it&apos;s focussed on making sure that AI <a href="https://carnegieendowment.org/emissary/2026/06/trump-ai-order-china-competition"><u>does not upset</u></a> the Communist Party&apos;s information control and censorship apparatus.&#xa0;</p><p>However, we think Chinese regulators will eventually reign in the release of very capable models. Very capable open-weight models will be used for all sorts of hacking, some of which will make the Chinese Communist Party look bad. The Party will not like that kind of political risk, so cyber safety regulation is coming.&#xa0;&#xa0;</p><p>But right now there is an opportunity for the US government to engage the Chinese government to encourage and shape this regulation to its advantage.&#xa0;</p><p>So as an overall strategy, what we&apos;ve laid out above makes sense: Laws that free up frontier labs to combat distillation collectively; tightening chip controls to make sure the US stays ahead of China; and engaging with Chinese regulators to prevent the spread of top-tier hacking capabilities to all and sundry.&#xa0;</p><p>Will this be what happens? Long-term strategic thinking is not a Trump admin forte, so we&apos;ll hope for the best, but prepare for the status quo.</p><h2 id="irans-perfectly-calibrated-cyber-attack">Iran&apos;s Perfectly Calibrated Cyber Attack</h2><p>A cyber attack on Minnesota water utilities coupled with a US federal government warning raises the possibility that Iranian hackers have begun targeting US critical infrastructure in earnest.&#xa0;</p><p>This week more than 30 Minnesota community water systems were targeted by a &quot;coordinated cyberattack&quot;, <a href="https://mn.gov/mnit/media/blog/?id=38-761869"><u>according to</u></a> the state&#x2019;s IT services. Disruptions were <a href="https://dysruptionhub.com/south-st-paul-water-cyber-incident/"><u>reported</u></a> in a <a href="https://dysruptionhub.com/plymouth-minnesota-water-cyberattack/"><u>number</u></a> of <a href="https://dysruptionhub.com/maple-plain-water-cyber-incident/"><u>cities</u></a>, but the water remained safe to drink. The worst impact reported was in the city of Braham where <a href="https://dysruptionhub.com/braham-minnesota-water-cyberattack/"><u>residents were asked</u></a> to limit their consumption of water for a few hours as the city was relying on supply in a single water tower.&#xa0;</p><p>The attack came less than a week after US federal government agencies <a href="https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-097a"><u>updated a warning</u></a> about the possibility of Iranian state-affiliated cyber actors targeting US critical infrastructure. The warning cites incidents since March in which multiple victims across several critical infrastructure sectors had been compromised and that some victims &quot;experienced operational disruption and financial loss&quot;.&#xa0;</p><p>There is no formal attribution of the Minnesota attacks, but security firm Tenable <a href="https://www.tenable.com/blog/coordinated-cyberattack-on-minnesota-water-utilities-what-you-need-to-know"><u>says the &quot;operational pattern&quot; is consistent</u></a> with the CyberAv3ngers, a threat actor associated with Iran&apos;s Islamic Revolutionary Guard Corp (IRGC).&#xa0;</p><p>In March we wrote that even if the war in Iran successfully achieved the White House&apos;s <a href="https://www.whitehouse.gov/articles/2026/03/operation-epic-fury-decisive-american-power-to-crush-irans-terror-regime/"><u>stated objectives</u></a>, it could still &quot;result in an enduring increase in Iran&apos;s capacity and appetite for cyber mayhem&quot;. Cyber operations targeting the US and Israel could provide propaganda victories with very little chance of a kinetic response.&#xa0;</p><p>The Minnesota attacks caused minimal impact, so it would be tempting to argue that they were a flop. Given that the war is <a href="https://www.semafor.com/newsletter/07/28/2026/semafor-gulf-five-months"><u>still ongoing</u></a> and <a href="https://www.natesilver.net/p/iran-war-polls-popularity-approval"><u>also very unpopular</u></a>, however, we think these attacks might be perfectly calibrated. They&apos;re significant enough to generate&#xa0; headlines and public unease, and, maybe, make Americans wonder about the point of the entire war. But they&apos;re not lethal or damaging enough to ensure America bands together against a single enemy.</p><p>Our crystal ball says we can expect more of this.&#xa0;</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/yv1zks-HXHo?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Chipping away at Chinese AI risks"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>US visa restrictions for cybercriminals:</strong> Last week, the US Secretary of State Marco Rubio <a href="https://www.state.gov/releases/office-of-the-spokesperson/2026/07/new-visa-restriction-policy-to-deter-and-dismantle-cyberscams-and-sextortion/"><u>announced</u></a> visa restrictions for cybercriminals including sextortionists and those running cyber-enabled scams. The policy applies to the &quot; individuals responsible&quot; for the crimes and, in some cases, their family members.&#xa0;</li><li><strong>AI security tools are a thing:</strong> This week Microsoft <a href="https://microsoft.ai/news/introducing-mai-cyber-1-flash-inside-mdash/"><u>announced a security-focussed AI model</u></a> that, when coupled with its <a href="https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/?msockid=07e1b320223d63fa1e34a47d23db623b"><u>MDASH harness</u></a>, is cheaper and better at vulnerability identification and remediation than current frontier models. It&apos;s not this specific announcement that makes us cheerful, but rather what it indicates. Microsoft announcing a &quot;world-class security&quot; machine as a product means that there will likely be a number of competitors offering a variety of similar products.&#xa0;&#xa0;</li><li><strong>AI works on cryptographic algorithms too: </strong>Anthropic has <a href="https://www.anthropic.com/research/discovering-cryptographic-weaknesses"><u>posted about Claude</u></a> Mythos Preview being used to find flaws in two different cryptographic systems. One of the results came in a proposed post-quantum algorithm that was being evaluated by NIST. This is a good example of AI being used preventatively to improve standards before they decided.&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, James Wilson chats with Airlock Digital co-founders David Cottingham and Daniel Schell about how attackers are using LLMs to enumerate EDR detections.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI137.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI137/&quot;&gt;Sponsored: How AI is putting pressure on EDR&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 17:58 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: How AI is putting pressure on EDR&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="shorts">Shorts</h2><h3 id="north-korean-hackers-hack-north-korea">North Korean Hackers Hack North Korea&#xa0;</h3><p>The <em>Daily NK</em> <a href="https://www.dailynk.com/english/north-korea-elite-bank-hacking-ring-arrested/"><u>reported last week</u></a> that a group of former North Korean military intelligence operatives had hacked two of the country&apos;s banks to steal funds for their own personal enrichment.&#xa0;</p><p>A source told the <em>Daily NK</em> that the group had stolen funds from two North Korean banks, the Chosun Central Bank and the Foreign Trade Bank. The stolen funds were then converted into cryptocurrency and laundered before being converted back into cash in China. The ringleaders were cyber operations veterans formerly from North Korea&apos;s <a href="https://en.wikipedia.org/wiki/Reconnaissance_General_Bureau"><u>Reconnaissance General Bureau</u></a>.&#xa0;</p><p>It looks like the outcome will be grim for the hackers. Per The Daily NK:</p><p>Officials in Pyongyang expect harsh punishment for those involved. &quot;They used the skills the state trained them with to defend the country, and instead robbed the country&#x2019;s coffers,&quot; one official said, according to the source. &quot;This goes beyond ordinary guilt-by-association penalties. It will be hard for the entire family line to survive.&quot;</p><p>Crikey.&#xa0;</p><h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss how important people are to cyber power and whether the rise of AI is changing that.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN176.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN176/&quot;&gt;Between Two Nerds: Cyber is people&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 30:12 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Cyber is people&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/EK9INfhlHDI?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Cyber is people"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>New Chinese cyber contractor identified:</strong> The cyber sleuths at Intrusion Truth have uncovered a new secretive Chinese IT company that appears to work as a cyber contractor and tool developer for Chinese state-sponsored hacking operations.</p><p>Online clues appear to suggest that Guangdong Chanming appears to have developed RedRelay (aka ORBWEAVER), an ORB network (aka proxy botnet) that was used by almost a dozen Chinese APT groups to hide the origin of their attacks.</p><p>According to Intrusion Truth, the company&apos;s customers allegedly include the Chinese People&apos;s Liberation Army and the Ministry of Public Security, which manages China&apos;s police forces.</p><p>[<a href="https://news.risky.biz/risky-bulletin-new-chinese-cyber-contractor-identified/"><u>more</u></a> on<em> Risky Bulletin</em>]</p><p><strong>A JSON RCE bug is about to rock the Java world:</strong> Threat actors are exploiting a vulnerability in Alibaba&apos;s Fastjson, one of the Java ecosystem&apos;s most popular libraries for working with JSON-formatted data.</p><p>Active exploitation began last week, a day after details about the security flaw were revealed by cybersecurity firm <a href="https://fearsoff.org/research/fastjson-1-2-83-rce"><u>FearsOff</u></a>.</p><p>The attacks, first spotted and documented by <a href="https://www.imperva.com/blog/imperva-customers-protected-against-cve-2026-16723-critical-fastjson-1-x-zero-day-rce/"><u>Imperva</u></a> and <a href="https://threatbook.io/blog/fastjson-rce-1.2.83-active-exploitation-detected-detection-mitigation"><u>ThreatBook</u></a>, target CVE-2026-16723, a vulnerability that can enable unauthenticated remote code execution attacks against Java projects that use the Fastjson library as a component.</p><p>[<a href="https://news.risky.biz/risky-bulletin-a-json-rce-bug-is-about-to-rock-the-java-world/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Western cyber agencies warn of Russian hacks of Zimbra servers:</strong> Cybersecurity and intelligence agencies from multiple Western countries have issued joint security advisories on Thursday warning of a major Russian hacking campaign that&apos;s targeting Zimbra email servers.</p><p>The attacks have been going on since last year. The zero-day, tracked as CVE-2025-66376, was patched in November but attacks have been traced back to at least July.</p><p>The zero-day itself is a stored cross-site scripting (XSS) bug that allows the attackers to load malicious code inside a Zimbra webmail client via the CSS @import feature. The malicious code would load a web tool called Ulej (Russian for Beehive) that could be used to harvest credentials, session tokens, backup 2FA codes, browser-saved passwords, and the contents of the victim&#x2019;s mailbox going back 90 days.</p><p>[<a href="https://news.risky.biz/risky-bulletin-western-cyber-agencies-warn-of-russian-hacks-of-zimbra-servers/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>
  127. Srsly Risky Biz: Knives Are Out For Open-Weight AI Models

    Thu, 23 Jul 2026 07:41:22 -0000

    <p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://thinkst.com/"><em><u>Thinkst</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher</em></p>
    <img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca-1.jpeg" alt="Srsly Risky Biz: Knives Are Out For Open-Weight AI Models"><p><em>Your weekly dose of Seriously Risky Business news is written by </em><a href="https://twitter.com/tomatospy"><em><u>Tom Uren</u></em></a><em> and edited by Patrick Gray and Amberleigh Jack. This week&apos;s edition is sponsored by </em><a href="https://thinkst.com/"><em><u>Thinkst</u></em></a><em>.</em></p><p><em>You can hear a podcast discussion of this newsletter by searching for &quot;Risky Business News&quot; in your podcatcher or subscribing via </em><a href="https://risky.biz/feeds/risky-business-news/"><em><u>this RSS feed</u></em></a><em>.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/SRB176.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/SRB176/&quot;&gt;Srsly Risky Biz: Knives are out for open-weight AI models&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 24:36 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Srsly Risky Biz: Knives are out for open-weight AI models&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca.jpeg" class="kg-image" alt="Srsly Risky Biz: Knives Are Out For Open-Weight AI Models" loading="lazy" width="2000" height="1500" srcset="https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w600/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca.jpeg 600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1000/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca.jpeg 1000w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/size/w1600/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca.jpeg 1600w, https://storage.ghost.io/c/16/e9/16e9a748-ca66-4b3c-8590-85537131f696/content/images/2026/07/data-src-image-c85e2cf7-76f5-4f2e-b16d-66e87a3adeca.jpeg 2048w" sizes="(min-width: 720px) 720px"><figcaption><span style="white-space: pre-wrap;">Photo by</span><a href="https://unsplash.com/@lureofadventure?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Ali Kazal</span></u></a><span style="white-space: pre-wrap;"> on</span><a href="https://unsplash.com/photos/black-handled-knife-on-brown-log-QKCo0sBAm58?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> <u><span class="underline" style="white-space: pre-wrap;">Unsplash</span></u></a></figcaption></figure><p>Both the American and Chinese governments have signalled they plan to rein in open-weight AI models.&#xa0;</p><p>The Trump administration seems certain to add some Chinese technology companies to the Entity List to protect investment in American frontier AI models. Meanwhile, Beijing is apparently weighing applying export restrictions on Chinese AI tech, including open-weight models.</p><p>Overnight, different US government officials issued a strong, coordinated signal that they plan to take action against Chinese AI companies. Michael Kratsios, the director of the White House&apos;s Office of Science and Technology Policy, <a href="https://x.com/mkratsios47/status/2079933645888880708?s=20"><u>wrote on X</u></a> that Chinese company Moonshot AI had &quot;developed a sophisticated internal platform to conduct large scale distillation against US models&quot; and that &quot;large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.&quot;</p><p>Treasury Secretary Scott Bessent also weighed in, <a href="https://x.com/SecScottBessent/status/2080008411790368895?s=20"><u>saying on X</u></a> that &quot;sanctions and Entity List designations will be on the table&quot;.&#xa0;</p><p>This has all come hot on the heels of two recent and interesting model launches out of China: GLM 5.2 from Z.ai and Moonshot AI&apos;s Kimi K3.&#xa0;</p><p>Moonshot <a href="https://www.kimi.com/blog/kimi-k3"><u>announced its new Kimi model</u></a> late last week and although it isn&apos;t open-weight yet, Moonshot says it will release the weights for Kimi K3 on 27 July.&#xa0;</p><p>According to independent model evaluations, it&apos;s very good. The Artificial Analysis Intelligence Index ranks the model <a href="https://x.com/ArtificialAnlys/status/2077832874183860404?s=20"><u>at number three</u></a>, the Vals AI index <a href="https://x.com/ValsAI/status/2077834614668988586"><u>ranks it number two</u></a>, and it ranks number one <a href="https://x.com/arena/status/2077824029126504525"><u>according to the Frontend Code Arena</u></a>.&#xa0;</p><p>Moonshot AI itself is more measured, saying it &quot;still trails the most powerful proprietary models&quot;.&#xa0;</p><p>No matter <em>exactly</em> how good Kimi K3 is, from the perspective of America&apos;s leading AI companies, having highly capable, open weight alternatives undermining your business model is A Bad Time. One source that <a href="https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi"><u><em>Axios</em> described as</u></a> &quot;close to the Trump administration&quot; says leading AI labs or their allies lobby the admin to ban open-weights models every 3-5 months.</p><p>They&apos;re furious about their models being distilled, but we&apos;re not convinced that distillation is the whole ballgame here.&#xa0;</p><p>Dean Ball, head of strategic future at OpenAI <a href="https://x.com/deanwball/status/2078133895766114412"><u>wrote on X</u></a> that he didn&apos;t think the performance of Kimi K3 &quot;can be explained away by distillation&quot;. Similarly, AI researcher and author of the <a href="https://substack.com/@natolambert"><u>Interconnects substack</u></a> Nathan Lambert, also thinks the <a href="https://x.com/natolambert/status/2080049053430390878?s=20"><u>impact of distillation is overstated</u></a> and that AI training pipelines are evolving so that distillation <a href="https://natolambert.substack.com/p/how-distillation-is-used-today-and"><u>is becoming less important</u></a> anyway.&#xa0;</p><p>Despite that, it seems that&#xa0; <a href="https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi?utm_campaign=article_email&amp;utm_content=article-17490"><u>internal debates</u></a> <a href="https://x.com/jacob_wendler/status/2075565694322651376?s=20"><u>within the Trump</u></a> administration <a href="https://www.semafor.com/article/07/15/2026/washington-confronts-chinas-open-source-models"><u>regarding open-weight Chinese models</u></a> have been settled: The Chinese are stealing American IP, and there needs to be a reckoning.&#xa0;</p><p>That&apos;s a shame, because these open-weight models are currently filling a gap not served by America&apos;s frontier labs.</p><p>Take <a href="https://huggingface.co/blog/security-incident-july-2026"><u>this month&apos;s hack</u></a> of US AI development company <a href="https://en.wikipedia.org/wiki/Hugging_Face"><u>Hugging Face</u></a> by rogue OpenAI models. This bonkers hack was discussed extensively on <a href="https://risky.biz/RB845/"><u>this week&apos;s episode</u></a> of the <em>Risky Business</em> podcast, but the short description is that some OpenAI models hacked the company&apos;s own infrastructure to escape a sandbox, then hacked Hugging Face, all in order to cheat on a cyber security evaluation test.</p><p>When Hugging Face tried to analyse the intrusion with frontier models it was stymied by safety guardrails. The analysis required sending real attack commands, exploit payloads and command and control artefacts to the frontier labs, which couldn&apos;t tell whether they might be helping an attacker instead of legitimate incident response.</p><p>So instead, Hugging Face turned to the Chinese open-weights GLM 5.2 model from Z.ai, running it on its own hardware. The company said:</p><blockquote>To understand what a swarm of tens of thousands of automated actions did, we ran LLM-driven analysis agents over the full attacker action log, comprised of more than 17,000 recorded events. This allowed us to reconstruct the timeline, extract indicators of compromise, map the credentials touched, and separate genuine impact from decoy activity. Thanks to this approach, we were able to do in hours what would usually take days, and match the adversary&apos;s speed.</blockquote><p>Using a self-hosted open-weight model also meant the company could guarantee that none of its sensitive data needed to be shipped off elsewhere. So American companies have three good reasons to turn to Chinese open-weight models: They&apos;re cheaper, you can run them on your own hardware, and they&apos;ve proven capable at cyber security tasks.</p><p>That&apos;s why we don&apos;t think it makes sense for the US government to outright ban foreign open-weights models.&#xa0;</p><p>The Chinese government has a say here too and there are <a href="https://www.reuters.com/world/beijing-is-looking-curbing-overseas-access-chinas-top-ai-models-sources-say-2026-07-07/"><u>emerging reports</u></a> that Beijing <a href="https://www.ft.com/content/6049a031-9e9b-464c-97bb-414da04d5a6a"><u>is considering</u></a> controlling the export of new open-weight models.&#xa0;</p><p>Until now, releasing open-weights models has made sense to Chinese companies and the Chinese government. Z.ai&apos;s Director of Product Zixuan Li <a href="https://www.chinatalk.media/p/the-zai-playbook"><u>told the <em>ChinaTalk</em> substack</u></a> in November last year that companies do it to contribute to research in the field, but also to build acceptance overseas.&#xa0;</p><p>&quot;I think it is necessary to be open right now for people to use our models&quot;.&#xa0;</p><p>From the Chinese government&apos;s perspective, it has <a href="https://www.reuters.com/world/asia-pacific/chinas-xi-promotes-chinas-commitment-ai-access-speech-shanghai-conference-2026-07-17/"><u>promoted the position that AI</u></a> is a global public good that should be developed for the benefit of all humankind. But this strategy is colliding with the reality that as these models become more capable they become far more dangerous.</p><p>We don&apos;t think it will be too long before open-weight models are capable of the sort of hacking in the Hugging Face example we described previously. This involved a combination of OpenAI models, including its current best GPT&#x2011;5.6 Sol and an even more capable pre-release model.&#xa0;</p><p>A report from the UK&apos;s AI Security Institute (AISI) <a href="https://www.aisi.gov.uk/blog/how-far-behind-the-frontier-are-leading-open-weight-models-on-cyber"><u>released last week</u></a> found that leading open-weight models were just four to seven months behind the frontier models, depending upon what you were measuring. That report examined Chinese open-weight models released in mid-April, with OpenAI&apos;s GPT-5.6 Sol and Claude Mythos 5. Kimi K3 appears to be a significant advance over those mid-April models.&#xa0;</p><p>Of course in the Hugging Face incident, OpenAI had turned off safeguards to get a better read on exactly how good its models were. Research <a href="https://www.lexology.com/library/detail.aspx?g=869c5f65-8f9f-4bc1-bbfd-332c9fbd95fd"><u>released in May this year</u></a> shows it is possible to remove safeguards from open-weights models quickly and painlessly. So as new and increasingly capable open-weights models are released, someone will remove safeguards so that they can be misused. And thanks to OpenAI&apos;s sloppy testing sandbox we&apos;ve seen what that&apos;ll look like.</p><p>So having its tech firms drop increasingly powerful models with easily removed safeguards onto the internet for Joe Public to use to cause chaos doesn&apos;t seem like a recipe that the Chinese government will be comfortable with for long. They&apos;re not exactly libertarians over there, after all.</p><p>The American and the Chinese governments each have their own motivations. But we suspect the current golden age of highly capable open-weight models dropping onto the internet every other week is coming to an end.</p><h2 id="for-scattered-spider-the-chickens-are-finally-coming-home">For Scattered Spider, the Chickens Are Finally Coming Home&#xa0;</h2><p>A string of successes in recent months suggests that law enforcement is finally getting a handle on the internet&apos;s brashest hackers: Teenagers.</p><p>Last week two members of the so-called Scattered Spider group <a href="https://www.bbc.com/news/articles/c4gyg0y6yg2o"><u>were each sentenced</u></a> in the UK to five and half years in prison for their roles in a 2024 breach of Transport for London&apos;s network. Thalha Jubair, 20 and Owen Flowers, 18, were described by prosecutors as leading members of Scattered Spider. In April this year, Tyler Robert Buchanan, 24, also <a href="https://krebsonsecurity.com/2026/04/scattered-spider-member-tylerb-pleads-guilty/"><u>pleaded guilty</u></a> to different set of Scattered Spider-related charges.&#xa0;</p><p>Early this month the US Department of Justice (DoJ) announced that another alleged member had <a href="https://www.justice.gov/opa/pr/alleged-member-criminal-cyber-hacking-group-scattered-spider-arrested-finland-and-extradited"><u>been arrested</u></a>. Peter Stokes, 19, a dual US-Estonian citizen, was detained by Finnish authorities in April and extradited to the US in late June. The criminal complaint alleges that Stokes was involved in a number of incidents, including the breaching and attempted ransom of a luxury jewelry retailer.&#xa0;</p><p>The DoJ says that Scattered Spider has been involved in over 100 network intrusions resulting in more than USD&#x24;100 million in ransom payments.&#xa0;</p><p>We first <a href="https://news.risky.biz/lapsus-from-flash-in-the-pan-to-raging-65c0eb42a0e6da001a37dfbe/"><u>covered the group</u></a> in 2023 in the wake of breaches at Caesars Entertainment and MGM Resorts International, two large US resort, entertainment and gaming companies. Scattered Spider shared some members with a previous group known as Lapsus&#x24; that had been so outrageously successful that it was the subject of a <a href="https://news.risky.biz/why-russias-cyber-war-against-ukraine-65c0eb42a0e6da001a37dfc2/#advanced-persistent-teenagers"><u>US Cyber Safety Review Board report</u></a> that year.&#xa0;</p><p><a href="https://cyberscoop.com/potent-youth-cybercrime-ring-made-up-of-1000-people-fbi-official-says/"><u>Reporting in 2024</u></a> described the group not so much as a discrete gang, but as an ecosystem or community that shares techniques and teaches juveniles to be effective hackers. Given that its members teamed up for particular projects, <a href="https://news.risky.biz/tiktok-manipulation-report-is-too-little-too-late/#scattered-spider-is-the-hollywood-of-cybercrime"><u>we likened it</u></a> to Hollywood as a collective, rather than a single production company.&#xa0;</p><p>In mid-2025, however, <a href="https://news.risky.biz/four-key-players-drive-scattered-spider/"><u>several security firms</u></a> said that a small number of key members were driving the activities of Scattered Spider, with different firms suggesting there were between two to four key players, who were essentially project managers. They selected targets and built a team of people from the broader group to carry out the attacks.&#xa0;</p><p>For key Scattered Spider members, elite social engineering skills appeared to be a key skill. Ransomware incident response firm Coveware <a href="https://www.coveware.com/blog/2023/10/27/scattered-ransomware-attribution-blurs-focus-on-ir-fundamentals"><u>wrote in 2023 that</u></a> &quot;skillful social engineering of the IT support desk to subvert, reset or overcome multi-factor authentication&quot; was a common tactic and that recordings confirmed that the same &quot;two or three&quot; individuals were consistently involved.&#xa0;&#xa0;</p><p>Although it has taken some time, these key individuals are now getting arrested. And it could just be because they have grown up.&#xa0;</p><p>In <a href="https://www.zetter-zeroday.com/tracking-peter-stokes-and-the-com-allison-nixon-and-her-work-unmasking-cybercriminals/"><u>an interview</u></a> with <em>Zero Day,</em> juvenile cybercrime expert Alison Nixon described why it is difficult to both deter and arrest young offenders involved with Scattered Spider&apos;s activities. Nixon says the FBI will not arrest juveniles because &quot;there is no federal juvenile system&quot;. It&apos;s only once they turn 18 and can be charged as an adult are suspects arrested. As for deterrence, Nixon says steering teenagers with no concept of the future onto the straight and narrow is difficult.</p><p>At this point, several influential members of Scattered Spider are aging out, have been arrested and some even sentenced to jail. We are hopeful that it will quiet the group, at least for a while.&#xa0;</p><p><em>Watch James Wilson and Tom Uren discuss this edition of the newsletter:</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/z2X9DLJkBZ4?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Srsly Risky Biz: Knives are out for open-weight AI models"></iframe></figure><h2 id="three-reasons-to-be-cheerful-this-week">Three Reasons to Be Cheerful This Week:</h2><ol><li><strong>Information Commissioner lets Qantas off the hook:</strong> The Office of the Australian information Commissioner <a href="https://www.oaic.gov.au/privacy/privacy-assessments-and-decisions/privacy-decisions/Investigation-inquiry-reports/report-into-preliminary-inquiries-of-qantas"><u>released its preliminary inquiry</u></a> into the June 2025 data breach at Australian airline Qantas. The OAIC doesn&apos;t plan to launch a full investigation because it thinks Qantas had taken &quot;reasonable&quot; steps to protect its data and had responded quickly to remediate the breach. The incident has all the hallmarks of a Scattered Spider attack, so the good news here is that it is possible to get owned by kids and yet somehow avoid the ire of the regulator.&#xa0;</li><li><strong>Kratos takedown:</strong> German and US authorities <a href="https://www.bka.de/DE/Presse/Listenseite_Pressemitteilungen/2026/Presse2026/260720_PM_Kratos.html"><u>have taken down</u></a> the <a href="https://any.run/cybersecurity-blog/kratos-phaas-account-takeover/"><u>Kratos phishing-as-a-service operation</u></a> and disrupted 200 servers. Its developer was also arrested by authorities in Indonesia.&#xa0;&#xa0;</li><li><strong><em>Krebs on Security</em> fixes LG TVs:</strong> LG has committed to ridding its webOS smart TV store of apps that turn a consumer&apos;s television into a residential proxy node. The promise came after <em>Krebs On Security</em> <a href="https://spur.us/blog/smart-tv-apps-residential-proxy-sdks"><u>highlighted research</u></a> that found 42% of webOS smart TV apps enrolled the devices into these networks. Residential proxies are often used for malicious purposes.&#xa0;&#xa0;&#xa0;&#xa0;</li></ol><h2 id="sponsor-section">Sponsor Section</h2><p><em>In this <strong>Risky Business sponsor interview</strong>, Casey Ellis chats with Haroon Meer from Thinkst about building companies customers don&#x2019;t hate. Haroon explains why Thinkst still offers Canary tokens for free and why it has avoided annual price hikes on its paid products. They talk about Eric Ries&#x2019;s &#x201c;Incorruptible&#x201d;, Rob Lee&#x2019;s 100-year-company approach at Dragos, and why keeping customers happy is a better business strategy than chasing easy sugar highs.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/RBNEWSSI136.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/RBNEWSSI136/&quot;&gt;Sponsored: Thinkst on building companies that don&#x2019;t suck&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 21:20 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Sponsored: Thinkst on building companies that don&#x2019;t suck&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <h2 id="risky-biz-talks">Risky Biz Talks</h2><p><em>You can find the audio edition of this newsletter and other fine podcasts and interviews in the Risky Biz News feed (</em><a href="https://risky.biz/feeds/risky-business-news"><em><u>RSS</u></em></a><em>, </em><a href="https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970"><em><u>iTunes</u></em></a><em> or </em><a href="https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q"><em><u>Spotify</u></em></a><em>).&#xa0;&#xa0;</em></p><p><em>In our last &quot;Between Two Nerds&quot; discussion Tom Uren and </em><a href="https://twitter.com/thegrugq"><em><u>The Grugq</u></em></a><em> discuss just how important exploits are for cyber operations using data published in a new paper authored by two members of Ukraine&#x2019;s cyber security agency.</em></p> <!--kg-card-begin: html--> <iframe frameborder="0" style="width: 100%; height: 156px;" srcdoc=" &lt;!DOCTYPE html&gt; &lt;html&gt; &lt;head&gt; &lt;title&gt;Risky Business Player&lt;/title&gt; &lt;/head&gt; &lt;body&gt; &lt;!-- Include the Google Font Inter --&gt; &lt;style&gt; @import url(&quot;https://fonts.googleapis.com/css2?family=Inter:wght@400;700&amp;display=swap&quot;); body { margin: 0px; } .audio-player { font-family: &quot;Inter&quot;, sans-serif; } .icon { background-image: url(&quot;https://risky.biz/static/img/icons/subscribe-icons.svg&quot;); display: block; width: 33px; height: 33px; background-size: 528px 111px; } } &lt;/style&gt; &lt;!-- Audio player --&gt; &lt;div class=&quot;audio-player&quot; style=&quot;display: flex; flex-direction: column; gap: 5px; padding-top: 0px; padding-bottom: 20px; background: rgb(244, 244, 239); background: linear-gradient(0deg, rgba(244, 244, 239, 1) 0%, rgba(244, 244, 239, 0) 100%); border-radius: 8px; border: 1px solid #d7d7d7; width: calc(100% - 1px) min-width: 350px;&quot;&gt; &lt;audio class=&quot;audioElement&quot; preload=&quot;none&quot;&gt; &lt;source src=&quot;https://dts.podtrac.com/redirect.mp3/media3.risky.biz/BTN174.mp3&quot; type=&quot;audio/mpeg&quot;&gt; Your browser does not support the audio element. &lt;/audio&gt; &lt;!-- Title --&gt; &lt;div style=&quot;background: #666666; color: #FEFEFE; font-size: 14px; padding-left: 5%; padding-right: 5%; padding-top: 10px; padding-bottom: 10px; border-top-left-radius: 8px; border-top-right-radius: 8px; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; margin-bottom: 10px;&quot;&gt; &lt;a style=&quot;color:#FEFEFE; text-decoration: none;&quot; target=&quot;_new&quot; href=&quot;https://risky.biz/BTN174/&quot;&gt;Between Two Nerds: Exploits are not cyber power&lt;/a&gt; &lt;/div&gt; &lt;!-- Player Controls and Progress Bar --&gt; &lt;div style=&quot;display: flex; align-items: center; justify-content: center; gap: 10px; width: 90%; margin: 0 auto;&quot;&gt; &lt;button type=&quot;button&quot; class=&quot;playPauseBtn&quot; style=&quot;color: #1e1e1e; background-color: #FAFAFA; font-size: 18px; border: none; padding: 10px; border-radius: 8px; cursor: pointer; height: 42px; width: 42px; text-align: center; display: flex; align-items: center; justify-content: center;&quot;&gt;&amp;#9654;&lt;/button&gt; &lt;input type=&quot;range&quot; style=&quot;flex-grow: 1; -webkit-appearance: none; height: 5px; background: #ddd; border-radius: 8px; outline: none; cursor: pointer;&quot; class=&quot;progressBar&quot; value=&quot;0&quot; min=&quot;0&quot; max=&quot;100&quot; /&gt; &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;currentTime&quot;&gt;0:00&lt;/span&gt; / &lt;span style=&quot;font-size: 12px; font-weight: 200;&quot; class=&quot;duration&quot;&gt; 30:08 &lt;/span&gt; &lt;/div&gt; &lt;!-- Subscribe Buttons --&gt; &lt;div style=&quot;width: 90%; display: flex; justify-content: space-between; align-items: center; padding-left: 5%;&quot;&gt; &lt;div style=&quot;padding-right:20px;&quot; class=&quot;subText&quot;&gt; &lt;strong&gt;Subscribe &amp;nbsp;&lt;/strong&gt; &lt;/div&gt; &lt;div style=&quot;display: flex; align-items: center; gap: 6px; margin: 0; flex-grow: 1;&quot; class=&quot;subContainer&quot;&gt; &lt;a href=&quot;https://podcasts.apple.com/au/podcast/risky-business-news/id1621305970&quot; class=&quot;icon apple-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 48px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://overcast.fm/itunes1621305970&quot; class=&quot;icon overcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 141px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://pca.st/yicebxgl&quot; class=&quot;icon pocketcast-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 234px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://open.spotify.com/show/0BdExoUZqbGsBYjt6QZl4Q&quot; class=&quot;icon spotify-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 420px 96px;&quot;&gt;&lt;/a&gt; &lt;a href=&quot;https://risky.biz/feeds/risky-business-news&quot; class=&quot;icon rss-podcast&quot; target=&quot;_new&quot; style=&quot;background-position: 327px 96px;&quot;&gt;&lt;/a&gt; &lt;/div&gt; &lt;a href=&quot;https://risky.biz&quot;&gt; &lt;img src=&quot;https://risky.biz/static/img/RB_Site_Logo.svg&quot; alt=&quot;Logo&quot; style=&quot;margin-left: 0; height: 32px; display: block; padding-right: 5%;&quot; id=&quot;logo&quot; class=&quot;logo playerLogo&quot;&gt; &lt;/a&gt; &lt;script&gt; document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); function resizeElements(player) { const logo = player.querySelector(&quot;.logo&quot;); const subscribeIcons = player.querySelectorAll(&quot;.icon&quot;); const subscribeContainer = player.querySelector(&quot;.subContainer&quot;); // Select subContainer by class const subText = player.querySelector(&quot;.subText&quot;); // Select subText by class if (player.offsetWidth &lt;= 425) { // Hide logo if (logo) { logo.style.display = &quot;none&quot;; } } else if (player.offsetWidth &lt;= 500) { // Show logo and scale logo and icons to 70% if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(0.7)&quot;; logo.style.transformOrigin = &quot;center&quot;; logo.style.verticalAlign = &quot;middle&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(0.7)&quot;; icon.style.transformOrigin = &quot;center&quot;; icon.style.verticalAlign = &quot;middle&quot;; }); // Remove padding from subText and set font-size to 12px if (subText) { subText.style.padding = &quot;0&quot;; subText.style.fontSize = &quot;12px&quot;; } // Set gap in subContainer to 0px if (subscribeContainer) { subscribeContainer.style.gap = &quot;0px&quot;; } } else { // Reset scaling, alignment, and visibility if (logo) { logo.style.display = &quot;block&quot;; logo.style.transform = &quot;scale(1)&quot;; logo.style.verticalAlign = &quot;baseline&quot;; } subscribeIcons.forEach(icon =&gt; { icon.style.transform = &quot;scale(1)&quot;; icon.style.verticalAlign = &quot;baseline&quot;; }); // Reset padding and font-size in subText if (subText) { subText.style.padding = &quot;0 20px&quot;; // Default padding subText.style.fontSize = &quot;inherit&quot;; // Default font-size } // Reset gap in subContainer if (subscribeContainer) { subscribeContainer.style.gap = &quot;6px&quot;; // Default gap } } } function handleResize() { players.forEach(player =&gt; { resizeElements(player); }); } // Run on initial load and resize handleResize(); window.addEventListener(&quot;resize&quot;, handleResize); }); &lt;/script&gt; &lt;/div&gt; &lt;/div&gt; &lt;script&gt; // Custom Audio Player document.addEventListener(&quot;DOMContentLoaded&quot;, function () { const players = document.querySelectorAll(&quot;.audio-player&quot;); players.forEach(function (player) { // Skip if already initialized if (player.dataset.initialized === &quot;true&quot;) return; // Mark player as initialized player.dataset.initialized = &quot;true&quot;; const audio = player.querySelector(&quot;.audioElement&quot;); const playPauseBtn = player.querySelector(&quot;.playPauseBtn&quot;); const progressBar = player.querySelector(&quot;.progressBar&quot;); const currentTimeEl = player.querySelector(&quot;.currentTime&quot;); const durationEl = player.querySelector(&quot;.duration&quot;); if (!audio || !playPauseBtn || !progressBar || !currentTimeEl || !durationEl) { console.error(&quot;One or more player elements not found:&quot;, { audio, playPauseBtn, progressBar, currentTimeEl, durationEl }); return; } playPauseBtn.addEventListener(&quot;click&quot;, () =&gt; { if (audio.paused) { audio.play(); playPauseBtn.textContent = &quot;&#x23f8;&quot;; // GA4 event for starting audio gtag(&quot;event&quot;, &quot;audio_play&quot;, { &quot;content_title&quot;: &quot;Between Two Nerds: Exploits are not cyber power&quot;, &quot;content_type&quot;: &quot;audio&quot; }); } else { audio.pause(); playPauseBtn.textContent = &quot;&#x25b6;&quot;; } }); audio.addEventListener(&quot;timeupdate&quot;, () =&gt; { if (audio.duration) { progressBar.value = (audio.currentTime / audio.duration) * 100; currentTimeEl.textContent = formatTime(audio.currentTime); } }); audio.addEventListener(&quot;loadedmetadata&quot;, () =&gt; { durationEl.textContent = formatTime(audio.duration); }); progressBar.addEventListener(&quot;input&quot;, () =&gt; { if (audio.duration) { audio.currentTime = (progressBar.value / 100) * audio.duration; } }); function formatTime(seconds) { const minutes = Math.floor(seconds / 60); const secs = Math.floor(seconds % 60); return `&#x24;{minutes}:&#x24;{secs &lt; 10 ? &quot;0&quot; : &quot;&quot;}&#x24;{secs}`; } }); }); &lt;/script&gt; &lt;/body&gt; &lt;/html&gt; "></iframe> <!--kg-card-end: html--> <p><em>Or watch it on YouTube!</em></p><figure class="kg-card kg-embed-card"><iframe width="200" height="113" src="https://www.youtube.com/embed/k6KALEsKpiA?feature=oembed" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen title="Between Two Nerds: Exploits are not cyber power"></iframe></figure><h2 id="from-risky-bulletin">From <a href="https://news.risky.biz/tag/risky-bulletin/"><em><u>Risky Bulletin</u></em></a>:</h2><p><strong>Linux kernel discloses 442 CVEs as AI bugpocalypse settles in:</strong> The Linux kernel project has disclosed <a href="https://lore.kernel.org/linux-cve-announce/"><u>442 vulnerabilities</u></a> over the past three days, in a massive dumb of CVEs on its security mailing list.</p><p>Although not confirmed, the bugs were likely discovered using AI tools. Over the past months, projects like Anthropic&apos;s Glasswing and OpenAI&apos;s Daybreak have been granting access to advanced frontier cybersecurity models to top-tier security firms and researchers to find bugs with AI in major open-source projects.</p><p>[<a href="https://news.risky.biz/risky-bulletin-linux-kernel-discloses-442-cves-as-ai-bugpocalypse-settles-in/"><u>more</u></a> on <em>Risky Bulletin</em>]</p><p><strong>Hacker wipes Romania&apos;s entire land registry database: </strong>A hacker has breached Romania&apos;s cadastre agency and wiped the country&apos;s entire land registry database following a failed extortion attempt.</p><p>The hack has brought Romania&apos;s entire real-estate market to a <a href="https://jurnaluldearges.ro/notarita-ana-stan-sunt-in-co-fortat-hackerii-au-spart-cadastrul-458711/"><u>standstill</u></a> as official apps and websites have been offline for a week. Notaries can&apos;t record new transactions while citizens can&apos;t obtain proof of ownership or detailed land records.</p><p>Email servers at the National Agency for Cadastre and Real Estate Advertising (<em>Agen&#x21b;ia Na&#x21b;ional&#x103; de Cadastru &#x219;i Publicitate Imobiliar&#x103;</em>, or ANCPI) were also down as part of the incident.</p><p>Sources told Risky Business that the hacker entered using valid credentials, mapped internal systems, and wiped systems and backups after failing to extort the agency.</p><p>[<a href="https://news.risky.biz/risky-bulletin-hacker-wipes-romanias-entire-land-registry-database/"><u>more</u></a> on <em>Risky Bulletin</em>]</p>