[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f4wx7DvPqREIMgrTzP79Q2LfBE6pcDylIR0JAjBdDw0w":3},{"roundup":4},{"id":5,"week_label":6,"slug":7,"date_from":8,"date_to":9,"tldr":10,"full_brief":11,"top_iocs":12,"social_linkedin":60,"social_x":61,"article_count":62,"awareness_links":63,"status":124,"published_at":125,"created_at":126,"updated_at":126,"mastodon_posted_at":127,"executive_summary":128,"tagline":129,"cover_image_url":130},"9e86cb67-b19e-461a-9b9c-febbd8211ac7","2026-W34","2026-w34","2026-08-17","2026-08-23","- 🔥 Microsoft Entra ID (CVSS 10.0) and Zimbra RCE flaws hit the CISA KEV catalog with confirmed in-the-wild exploitation this week\n- 🦀 North Korean hackers (Sapphire Sleet) poisoned popular Rust crates with 245M+ downloads, executing infostealers at build time\n- 🏦 Banking trojans Manic, Grandoreiro, and ToxicPanda 2.0 resurface with upgraded evasion, Wi-Fi mesh exfiltration, and global targeting\n- 🤖 AI-generated exploit scripts are actively targeting Siemens S7 PLCs in US critical infrastructure, lowering the attack bar significantly\n- 🔗 Supply chain risk exploded across npm, Rust, and Android car head units as attackers pivot to trusted update and build mechanisms\n- 🇪🇺 European regulators issued over €50M in GDPR fines this week alone, targeting telcos, data brokers, and publishers for security and consent failures\n- 🛡️ Ungoverned AI agents and \"shady AI\" governance gaps are producing real Sev-1 incidents, with no patch available for the process problem","## Vulnerabilities & Exploits\n\n**[Microsoft Entra ID Flaw (CVSS 10.0) Exploited in Wild, Allows Remote Code Execution](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fmicrosoft-entra-id-flaw-cvss-100.html)**. CVE-2026-69836, a maximum-severity RCE vulnerability in Microsoft Entra ID, has been actively exploited in the wild. Microsoft states the vulnerability is fully server-side mitigated and no customer action is required, but the absence of technical detail and the identity-layer targeting make this worth validating in your environment.\n\n**[CISA Orders Feds to Patch Actively Exploited TrueConf Server Flaws](https:\u002F\u002Fwww.bleepingcomputer.com\u002Fnews\u002Fsecurity\u002Fcisa-orders-feds-to-patch-actively-exploited-trueconf-server-flaws\u002F)**. Two critical TrueConf Server vulnerabilities, CVE-2026-72529 (missing authentication) and CVE-2026-72530 (code injection), are being actively exploited by the Head Mare hacktivist group to deploy PhantomCore backdoor malware via trojanized client installers. CISA has added both to its KEV catalog and mandated federal remediation.\n\n**[Attackers Exploit Zimbra SNMP Flaw for Unauthenticated RCE](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fattackers-exploit-zimbra-snmp-flaw-for.html)**. CVE-2026-73570, an OS command injection flaw in Zimbra Collaboration, is being actively exploited in the wild when the optional zimbra-snmp package is installed. CERT Polska has issued alerts and the flaw is now on the CISA KEV catalog; patch to version 10.1.20 immediately. [Learn more](\u002Fawareness\u002Funpatched-zimbra-flaw-enables-unauthenticated-remote-code-execution)\n\n**[ShieldBreak Zero-Day: No Patch, CISA BOD Gives 14 Days](https:\u002F\u002Fblog.qualys.com\u002Fproduct-tech\u002F2026\u002F08\u002F20\u002Fshieldbreak-the-windows-defender-zero-day-with-no-patch-detect-it-mitigate-it-with-qualys)**. CVE-2026-69414 (ShieldBreak) is an unpatched local privilege escalation to SYSTEM in Microsoft's Malware Protection Engine. A public PoC dropped August 12 and no patch exists yet, making detection and compensating controls the only option for defenders right now. [Learn more](\u002Fawareness\u002Fshieldbreak-zero-day-exposes-system-level-risk-with-no-patch-in-sight)\n\n**[Critical NetScaler Flaw Can Bypass Authentication on Gateway and AAA Servers](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fcritical-netscaler-flaw-can-bypass.html)**. Citrix patched CVE-2026-19490 (CVSS 9.3), an authentication bypass affecting NetScaler ADC and Gateway appliances configured with SAML actions, alongside a memory overflow DoS flaw. Citrix is urging immediate upgrades given the sensitivity of these perimeter devices to credential theft and lateral movement. [Learn more](\u002Fawareness\u002Fcitrix-netscaler-flaws-highlight-urgency-of-timely-patching-for-network-appliances)\n\n**[MLflow SSRF Vulnerability Actively Exploited for Cloud Credential Theft](https:\u002F\u002Fwww.securityweek.com\u002Fmlflow-vulnerability-exploited-for-cloud-credential-theft\u002F)**. CVE-2026-64849 (CVSS 9.3) in MLflow's unauthenticated model-registry webhooks API is being exploited to steal cloud credentials and secrets from hosted AI\u002FML instances. CISA added it to the KEV catalog; all MLflow versions before 3.15.0 are affected. [Learn more](\u002Fawareness\u002Fmlflow-ssrf-flaw-actively-exploited-to-steal-cloud-credentials)\n\n### Key Takeaway\nPrioritize patching Entra ID, TrueConf, Zimbra, NetScaler, and MLflow this week; all have confirmed exploitation and at least one carries a CVSS 10.0 score.\n\n---\n\n## Supply Chain\n\n**[Rust Supply Chain Attack Linked to North Korean Hackers](https:\u002F\u002Fwww.securityweek.com\u002Frust-supply-chain-attack-linked-to-north-korean-hackers\u002F)**. Sapphire Sleet compromised a maintainer account for the Rust crate `arrayref` and two others (`internment`, `append-only-vec`), injecting a malicious `proc-macro1` dependency that downloaded and executed an infostealer during the Cargo build process. The combined crates have over 245 million downloads, and the malware exfiltrated browser credentials and established persistence on developer workstations and CI\u002FCD systems.\n\n**[14 Trojanized npm Packages Drop RedC2 4.0 Linux Backdoor With AI-Assisted C2](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002F14-trojanized-npm-packages-drop-redc2.html)**. Fourteen malicious npm packages disguised as calendar and streak utilities were found delivering RedC2 4.0, an AI-powered Linux backdoor sold by threat actor MarlboroMan. The framework provides terminal access, file transfer, and in-memory execution capabilities, representing a significant escalation in AI-assisted post-exploitation tooling available to criminal actors.\n\n**[Android Car Malware Spreads Through Built-In Updaters](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fandroid-car-malware-spreads-through.html)**. The MoYu Group, previously linked to the BADBOX botnet, has weaponized legitimate update mechanisms in Android-based vehicle head units to deploy a multi-stage downloader for ad fraud and proxy botnet recruitment. This is the first documented malware supply chain attack specifically targeting automotive head units, raising new concerns about connected vehicle security. IOC: SHA256 `a7a70777175420766027f01658644645337090057b5462248176550316c7f29c` (TWCore update app used as dropper vehicle).\n\n### Key Takeaway\nAudit your build pipelines and developer dependency lockfiles immediately; build-time attacks executing at `cargo build` or `npm install` bypass most runtime security controls entirely.\n\n---\n\n## Ransomware & Breaches\n\n**[Apollo Global Management Discloses Social Engineering Breach](https:\u002F\u002Fcyberscoop.com\u002Fapollo-discloses-data-breach-social-engineering-attack\u002F)**. Private equity giant Apollo Global Management confirmed a breach between July 6-10 in which the BlackFile threat group (affiliated with The Com) used social engineering to compromise cloud platforms and access sensitive PII including SSNs, dates of birth, and home addresses. The campaign is part of an ongoing wave targeting financial and professional services firms.\n\n**[SickKids Data Breach Exposes Employee and Applicant Information](https:\u002F\u002Fwww.bleepingcomputer.com\u002Fnews\u002Fsecurity\u002Fsickkids-data-breach-exposes-employee-and-job-applicant-info\u002F)**. Toronto's Hospital for Sick Children disclosed a breach originating from a vulnerability in third-party software, exposing current and former employee and job applicant data. Clinical systems and patient records were unaffected, but this is the second major incident for the organization following a 2022 ransomware attack, underscoring the persistent risk from third-party software in healthcare.\n\n**[Hundreds of Leaked AWS Keys Give Full Control Over Corporate Accounts](https:\u002F\u002Fwww.bleepingcomputer.com\u002Fnews\u002Fsecurity\u002Fhundreds-of-leaked-aws-keys-give-full-control-over-corporate-accounts\u002F)**. Truffle Security identified over 9,300 active AWS access keys with administrative privileges publicly exposed in code repositories and other sources, with Hugging Face flagged as a significant exposure point. Attackers with these keys gain full account control, enabling data theft, server compromise, and cryptomining with no initial vulnerability required.\n\n### Key Takeaway\nSocial engineering and credential exposure remain the most reliable initial access vectors; enforce phishing-resistant MFA and rotate any secrets that have ever touched a public repository.\n\n---\n\n## APT & Nation-State\n\n**[Suspected Russian Hackers Abuse Google OAuth and WhatsApp Linking to Hijack Accounts](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fsuspected-russian-hackers-abuse-google.html)**. Three Russian espionage clusters (UNC6293\u002FAPT29, UNC7005, UNC5976) are targeting academia, aerospace, defense, and government in Europe and the US by abusing legitimate Google OAuth and WhatsApp authentication flows. UNC7005 deploys a rogue Excel plugin called HEADRUSH, while UNC5976 automates token collection at scale via cloud infrastructure and fake file-sharing lures.\n\n**[China's Volt Typhoon Is Pre-Positioning Malware in US Critical Infrastructure](https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fchina-is-strapping-digital-bombs-to-civilian-infrastructure-is-the-us-ready\u002F)**. A war game simulating a Chinese cyberattack on US water utilities demonstrated the real impact of Volt Typhoon's multi-year pre-positioning campaign inside critical infrastructure including water, power, and telecommunications systems. The objective is disruption and coercion, not espionage, representing a fundamentally different threat model that most defenders are not positioned to counter.\n\n**[AI-Generated Exploit Scripts Target Siemens S7 PLCs in US Critical Infrastructure](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fai-generated-exploit-scripts-target.html)**. The US government warned of active threat actors using AI-generated exploit scripts to target Siemens S7 PLCs via internet-scanning services. The use of AI dramatically lowers the technical barrier for ICS attacks, enabling credential access, reconnaissance, and denial of service against industrial processes that were previously accessible only to sophisticated actors.\n\n**[Pakistan's Transparent Tribe Refreshes Toolset for Afghan Cyberattacks](https:\u002F\u002Fwww.darkreading.com\u002Fcyberattacks-data-breaches\u002Fpakistan-transparent-tribe-afghan-cyberattacks)**. Transparent Tribe has updated its toolkit and shifted focus toward less mature Afghan organizations, particularly Taliban-affiliated bodies, after failing to penetrate more hardened Indian government targets. The campaign illustrates how nation-state actors calibrate their capabilities against target security maturity. [Learn more](\u002Fawareness\u002Fnation-state-actor-exploits-security-maturity-gaps-in-afghan-organizations)\n\n### Key Takeaway\nOT and ICS environments must be treated as active targets, not passive bystanders; network-segment PLCs, audit internet-facing industrial devices, and assume pre-positioned adversary presence in critical infrastructure.\n\n---\n\n## Malware & Emerging Threats\n\n**[Banking Trojans Manic, Grandoreiro, and ToxicPanda 2.0 in the Spotlight](https:\u002F\u002Fwww.securityweek.com\u002Fbanking-trojans-manic-grandoreiro-toxicpanda-2-0-in-the-spotlight\u002F)**. Three banking trojans are making headlines simultaneously: Manic (Android) uses a Wi-Fi mesh relay to exfiltrate data from offline devices through nearby infected ones; Grandoreiro resurfaces post-takedown with advanced anti-analysis features targeting Latin America and Europe; and ToxicPanda 2.0 expands command support and is now distributed via AWS. IOC: malware family `Manic`. [Learn more](\u002Fawareness\u002Fgrandoreiro-banking-trojan-returns-with-stronger-evasion-capabilities)\n\n**[Microsoft Defender's Own Driver Can Be Weaponized to Delete Security Software at Boot](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fmicrosoft-defenders-own-driver-can-be.html)**. Check Point Research found that Microsoft Defender's legitimate `BTR.sys` driver can be weaponized by an administrator-level attacker to delete security software, including Defender itself, during the boot process before user-mode services initialize. This is an architectural trust-boundary issue, not a software flaw, meaning no patch is forthcoming and detection must happen at the behavioral layer.\n\n**[New Phishing Toolkit Uses Passkeys to Maintain Access After Password Resets](https:\u002F\u002Fwww.securityweek.com\u002Fnew-phishing-toolkit-uses-passkeys-to-maintain-access-after-password-resets\u002F)**. The iAuthFlow V2 phishing toolkit, sold for $10,000, registers attacker-controlled passkeys to victim accounts during a phishing session, maintaining persistent access even after password resets and session revocations. This directly undermines passkey rollout as a security control if defenders do not also monitor for unauthorized passkey registration events.\n\n### Key Takeaway\nMonitor for anomalous passkey registrations and unauthorized FIDO2 credential additions in identity logs; a compromised session can now survive a full credential reset.\n\n---\n\n## Regulatory & Compliance\n\n**[French Regulators Issue Over €48M in GDPR Fines in a Single Week](https:\u002F\u002Fwww.edpb.europa.eu\u002Fnews\u002Fdata-breach-free-mobile-and-free-fined-eur42-million_en)**. FREE MOBILE and FREE were fined a combined €42 million for a breach exposing 24 million subscribers including IBANs, NEXPUBLICA FRANCE was fined €1.7 million for inadequate security in CRM software, and Mobius Solutions was fined €1 million for retaining and misusing 46 million Deezer user records after contract expiry. France Travail received a separate €5 million fine for a social engineering breach caused by inadequate authentication and logging controls.\n\n**[Cisco Patches Five CVSS 10.0 Flaws in Crosswork and Secure Workload](https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fcisco-patches-nine-crosswork-and-secure.html)**. Cisco released patches for nine vulnerabilities in its Crosswork and Secure Workload products, five of which scored CVSS 10.0 including SQL injection and missing authentication flaws. No active exploitation has been observed, but CVSS 10.0 ratings on network management and workload security platforms demand immediate attention. [Learn more](\u002Fawareness\u002Fcisco-patches-critical-flaws-enabling-rce-and-auth-bypass-in-crosswork-secure-workload)\n\n**[Atlassian and Splunk Patch 250+ Vulnerabilities, Many from Third-Party Libraries](https:\u002F\u002Fwww.securityweek.com\u002Fatlassian-splunk-patch-dozens-of-critical-high-severity-vulnerabilities\u002F)**. Atlassian addressed 172 issues across Bamboo, Bitbucket, Confluence, and Jira, while Splunk patched 150+ in Enterprise, SOAR, and Universal Forwarder. The majority stem from third-party library dependencies, reinforcing that software composition risk is now a first-order patching challenge. [Learn more](\u002Fawareness\u002Fatlassian-splunk-release-patches-for-250-vulnerabilities-many-from-third-party-libraries)\n\n### Key Takeaway\nRegulators are enforcing security adequacy requirements with nine-figure fines; document your patch timelines, authentication controls, and processor agreements now rather than after a breach notification.\n\n---\n\n## References\n\n- https:\u002F\u002Fwww.bleepingcomputer.com\u002Fnews\u002Fsecurity\u002Fcisa-orders-feds-to-patch-actively-exploited-trueconf-server-flaws\u002F\n- https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fmicrosoft-entra-id-flaw-cvss-100.html\n- https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Frust-supply-chain-attack-puts-build.html\n- https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fai-generated-exploit-scripts-target.html\n- https:\u002F\u002Fthehackernews.com\u002F2026\u002F08\u002Fmanic-android-malware-exfiltrates-data.html\n- https:\u002F\u002Fwww.wired.com\u002Fstory\u002Fchina-is-strapping-digital-bombs-to-civilian-infrastructure-is-the-us-ready\u002F\n- https:\u002F\u002Fwww.edpb.europa.eu\u002Fnews\u002Fdata-breach-free-mobile-and-free-fined-eur42-million_en\n- https:\u002F\u002Fwww.securityweek.com\u002Fmlflow-vulnerability-exploited-for-cloud-credential-theft\u002F",[13,17,20,23,26,30,34,37,40,43,45,48,50,54,57],{"type":14,"value":15,"context":16},"cve","CVE-2026-73570","Zimbra Collaboration vulnerability being actively exploited.",{"type":14,"value":18,"context":19},"CVE-2026-72529","TrueConf Server Missing Authentication for Critical Function Vulnerability",{"type":14,"value":21,"context":22},"CVE-2026-72530","TrueConf Server Code Injection Vulnerability",{"type":14,"value":24,"context":25},"CVE-2026-69836","Microsoft Entra ID vulnerability",{"type":27,"value":28,"context":29},"domain","cardoor[.]cn","MQTT message broker domain used for distributing malware updates.",{"type":31,"value":32,"context":33},"malware","Manic","Name of the Android malware family",{"type":31,"value":35,"context":36},"infostealer","Type of malware deployed",{"type":27,"value":38,"context":39},"23.254.165.112","C2 server IP address used by the loader.",{"type":27,"value":41,"context":42},"rasGThauFD.com","DGA hunting indicator for August 20-24, 2026.",{"type":27,"value":44,"context":42},"feVVKIiEiU.com",{"type":31,"value":46,"context":47},"snap7.dll","Custom Python script uses this library to mimic legitimate monitoring utilities.",{"type":31,"value":49,"context":47},"python-snap7",{"type":51,"value":52,"context":53},"hash_sha256","a7a70777175420766027f01658644645337090057b5462248176550316c7f29c","Hash of TWCore, the legitimate software used to distribute JarService.",{"type":51,"value":55,"context":56},"0004165580107613384171326934363878323834353339333331343430323430","Hash of Stage 1: JarService dropper.",{"type":51,"value":58,"context":59},"0445451111244211411145114411111111111111111111111111111111111111","Hash of Stage 2: loader.","This week in threat intelligence: attackers hit identity at CVSS 10.0, poisoned developer build tools, and aimed AI-generated exploits at industrial control systems.\n\nKey stories from the ThreatNoir 2026-W34 roundup:\n\n- Microsoft Entra ID (CVSS 10.0) and Zimbra RCE confirmed exploited and added to CISA KEV\n- North Korean Sapphire Sleet poisoned Rust crates with 245M+ downloads via build-time infostealer\n- AI-generated scripts actively targeting Siemens S7 PLCs in US critical infrastructure\n- iAuthFlow V2 phishing toolkit registers attacker passkeys to survive password resets\n- European regulators issued over 50M euros in GDPR fines in a single week\n\nFull roundup: https:\u002F\u002Fthreatnoir.com\u002Fweekly\u002F2026-w34\n\n#ThreatIntelligence #CyberSecurity #CISO #SupplyChainSecurity #Infosec","Week 34: Entra ID CVSS 10.0 exploited, Rust build-time infostealer tied to North Korea, AI scripts hitting Siemens PLCs, passkey persistence bypasses password resets. Patch list is long. Full roundup: https:\u002F\u002Fthreatnoir.com\u002Fweekly\u002F2026-w34",80,[64,67,70,73,76,79,82,85,88,91,94,97,100,103,106,109,112,115,118,121],{"slug":65,"title":66},"italian-dpa-fines-publisher-280k-for-marketing-emails-without-valid-consent","Italian DPA Fines Publisher €280K for Marketing Emails Without Valid Consent",{"slug":68,"title":69},"cisco-patches-critical-flaws-enabling-rce-and-auth-bypass-in-crosswork-secure-workload","Cisco Patches Critical Flaws Enabling RCE and Auth Bypass in Crosswork & Secure Workload",{"slug":71,"title":72},"ai-powered-tools-shore-up-satellite-network-after-russian-cyberattack","AI-Powered Tools Shore Up Satellite Network After Russian Cyberattack",{"slug":74,"title":75},"critical-mlflow-ssrf-vulnerability-actively-exploited-in-the-wild","Critical MLflow SSRF Vulnerability Actively Exploited in the Wild",{"slug":77,"title":78},"should-ai-be-designated-critical-infrastructure","Should AI Be Designated Critical Infrastructure?",{"slug":80,"title":81},"grandoreiro-banking-trojan-returns-with-stronger-evasion-capabilities","Grandoreiro Banking Trojan Returns With Stronger Evasion Capabilities",{"slug":83,"title":84},"14000-ip-cameras-compromised-via-chained-cves-and-weak-credentials","14,000 IP Cameras Compromised via Chained CVEs and Weak Credentials",{"slug":86,"title":87},"atlassian-splunk-release-patches-for-250-vulnerabilities-many-from-third-party-libraries","Atlassian & Splunk Release Patches for 250+ Vulnerabilities, Many from Third-Party Libraries",{"slug":89,"title":90},"citrix-netscaler-flaws-highlight-urgency-of-timely-patching-for-network-appliances","Citrix NetScaler Flaws Highlight Urgency of Timely Patching for Network Appliances",{"slug":92,"title":93},"mlflow-ssrf-flaw-actively-exploited-to-steal-cloud-credentials","MLflow SSRF Flaw Actively Exploited to Steal Cloud Credentials",{"slug":95,"title":96},"expired-visa-cards-can-be-revived-for-contactless-payments-via-nfc-manipulation","Expired Visa Cards Can Be Revived for Contactless Payments via NFC Manipulation",{"slug":98,"title":99},"italian-publisher-fined-280k-for-marketing-emails-sent-without-valid-gdpr-consent","Italian Publisher Fined €280K for Marketing Emails Sent Without Valid GDPR Consent",{"slug":101,"title":102},"ungoverned-ai-agents-create-serious-data-exposure-risks","Ungoverned AI Agents Create Serious Data Exposure Risks",{"slug":104,"title":105},"http3-to-http11-translation-flaw-enables-350x-dos-amplification-via-cdns","HTTP\u002F3-to-HTTP\u002F1.1 Translation Flaw Enables 350x DoS Amplification via CDNs",{"slug":107,"title":108},"manic-android-malware-uses-wi-fi-mesh-to-exfiltrate-data-from-offline-devices","Manic Android Malware Uses Wi-Fi Mesh to Exfiltrate Data from Offline Devices",{"slug":110,"title":111},"nation-state-actor-exploits-security-maturity-gaps-in-afghan-organizations","Nation-State Actor Exploits Security Maturity Gaps in Afghan Organizations",{"slug":113,"title":114},"unpatched-zimbra-flaw-enables-unauthenticated-remote-code-execution","Unpatched Zimbra Flaw Enables Unauthenticated Remote Code Execution",{"slug":116,"title":117},"critical-elementor-pro-file-upload-flaw-opens-wordpress-sites-to-rce","Critical Elementor Pro File Upload Flaw Opens WordPress Sites to RCE",{"slug":119,"title":120},"shieldbreak-zero-day-exposes-system-level-risk-with-no-patch-in-sight","ShieldBreak Zero-Day Exposes SYSTEM-Level Risk With No Patch in Sight",{"slug":122,"title":123},"ai-powered-phishing-outsmarts-traditional-email-filters","AI-Powered Phishing Outsmarts Traditional Email Filters","published","2026-08-23T05:00:02.695+00:00","2026-08-23T05:02:17.170813+00:00","2026-08-23T05:15:03.16+00:00","### The week in one line\nExploited CVSS 10.0 flaws, AI-accelerated attacks on industrial systems, and poisoned build tools define a week of compounding risk.\n\n### What happened\nAttackers exploited maximum-severity vulnerabilities across identity, messaging, and email platforms while simultaneously targeting developer supply chains and industrial control systems. The exploitation window continued to shrink, with GitLab CVE-2026-19478 weaponized within days of disclosure and AI tooling enabling faster script generation against ICS targets.\n\n- Microsoft Entra ID CVE-2026-69836 (CVSS 10.0) confirmed exploited in the wild\n- CISA added TrueConf CVE-2026-72529 and CVE-2026-72530 and Zimbra CVE-2026-73570 to the KEV catalog\n- North Korean Sapphire Sleet poisoned Rust crates with 245M+ downloads via a compromised maintainer account\n- AI-generated exploit scripts actively targeting Siemens S7 PLCs in US critical infrastructure\n- Apollo Global Management breached via social engineering by The Com-affiliated BlackFile group\n\n### Why it matters for defenders and leaders\nThe convergence of AI-assisted exploitation, supply chain compromise at the build layer, and identity platform attacks means that traditional perimeter and signature-based defenses are failing at multiple simultaneous points. CVSS 10.0 flaws in identity infrastructure are no longer theoretical; they are being weaponized in hours, not weeks.\n\n- Build-time supply chain attacks bypass runtime EDR and execute before any monitoring agent starts\n- Passkey-based persistence in iAuthFlow V2 survives password resets and session revocations\n- Pre-positioned Volt Typhoon implants in critical infrastructure represent a deterrence threat, not just an espionage risk\n- Ungoverned AI agents are producing real data exposure incidents with no software patch to apply\n\n### What to do this week\n- Patch or verify server-side mitigation for CVE-2026-69836 (Entra ID), CVE-2026-72529 and CVE-2026-72530 (TrueConf), CVE-2026-73570 (Zimbra), and CVE-2026-64849 (MLflow) per CISA KEV deadlines\n- Audit Rust and npm dependency lockfiles for `proc-macro1` and the 14 malicious calendar\u002Fstreak packages; review CI\u002FCD build logs for anomalous network calls\n- Enable alerting on new passkey and FIDO2 credential registrations across all identity platforms\n- Network-segment internet-facing PLCs and audit Siemens S7 devices for unauthorized access or configuration changes\n- Review all active AI agent permissions and data-access scopes; enforce least-privilege and require human approval for external data publishing","Build time is attack time, patch faster","https:\u002F\u002Fcdn.threatnoir.com\u002Fweekly\u002F2026-w34-cover.png"]