Weekly Threat Intel Report — 2026-W46: 9-15 November 2026
TL;DR
Three threads defined the week. Agentic AI penetration testing frameworks were used for live intrusions, with the FBI issuing a flash alert on the open-source Artex framework and Unit 42 documenting attacker use of Cairn against semiconductor, defense, and civil society targets. Intrusions progressed from initial targeting to privileged access in under three hours. Infrastructure vulnerabilities drove active exploitation: Check Point Security Gateways were abused via PHP2Shell to deploy VMware ESXi wipers, Check Point patched a separate zero-day granting root access that had been exploited since late October, and N-able took N-central offline following disclosure of 33 flaws, with the 4EverAble2Shell chain exploited within hours of PoC release. Apache HTTP Server patched a critical HTTP/2 RCE and Microsoft shipped fixes for 1,328 CVEs including 60 critical. Named-actor reporting covered Turla's new Cereal toolset with an Exchange mail collection module, KrebsOnSecurity's attribution of the Qilin administrator to a St. Petersburg resident, and Recorded Future's tracking of the DoppelCart fraud network at 260,000+ domains leveraging Medusa infrastructure. The U.S. disrupted a DPRK social engineering campaign responsible for $400 million in cryptocurrency theft. Cursor confirmed a marketplace breach that distributed a malicious Claude Code extension after Fireworks AI credential theft.
Notable Activity by Actor
Turla. ESET published a technical report on Turla's Cereal toolset. Cereal includes a Microsoft Exchange email collection module, a novel C2 mechanism, and .NET execution tricks that complicate analysis. The Exchange module targets on-premises mail servers for direct mailbox harvesting rather than relying on compromised user credentials. Detection content should focus on anomalous .NET assembly loading inside Exchange worker processes and the identified C2 traffic patterns in the ESET writeup. Separately, DarkReading cited research in which analysts used AI-assisted reverse engineering to unpack and recreate portions of Turla malware, indicating the tradecraft bar for mapping complex implants is dropping.
Qilin. KrebsOnSecurity named a 36-year-old resident of St. Petersburg as the administrator of Qilin, citing a decade of operational security errors across forum handles, payment records, and infrastructure. Qilin claimed at least 365 victim organizations in the past three months, making it the most active ransomware affiliate program this quarter. Attribution reporting of this kind does not disrupt operations by itself, but it constrains the actor's ability to travel and may feed future law enforcement action.
Medusa. Recorded Future tracked DoppelCart, a shopping fraud network that has grown to more than 260,000 domains impersonating global brands. Infrastructure analysis shows overlaps with Medusa operations, pointing to shared hosting and registration tradecraft across what are typically treated as distinct criminal clusters.
Luna Moth (standalone). DarkReading reported Luna Moth returning to law firms it had previously extorted, this time using social engineering aimed at stealing Microsoft access tokens rather than deploying encryption. The focus on tokens over passwords sidesteps MFA for users whose tokens are exfiltrated post-authentication.
CopperDragon (standalone). Check Point Research described the Sparrow framework and CopperDragon cluster, tying together intrusions across Latin America, Asia, and Europe that had been tracked as separate operations. The connective tissue is shared post-exploitation tooling and infrastructure overlap.
TAG-218 / GhostFrame / Settra (standalone). Recorded Future detailed GhostFrame, a post-exploitation framework targeting VMware ESXi hosts, linked to TAG-218 operations and Settra ransomware deployment. ESXi-resident implants continue to be a preferred persistence location because many organizations lack endpoint telemetry on hypervisors.
DPRK cryptocurrency theft. The U.S. government disrupted a DPRK social engineering campaign tracked as FaceStealer, responsible for more than $400 million in cryptocurrency theft. The disruption targeted infrastructure used for persona creation and lure delivery against exchange and wallet staff.
Emerging Threats
Agentic AI pentesting frameworks in the wild. The FBI issued a flash alert on Artex, an open-source AI pentesting framework now observed in live intrusions. Cisco Talos, Sophos, and Recorded Future all published parallel analyses. The reported effect is intrusions progressing from initial targeting to privileged access in under three hours, with 45 forks and related frameworks retaining the same capabilities. Unit 42 described a parallel case using the Cairn framework to autonomously breach semiconductor, defense, and civil society targets. A DarkReading case study walked through an AI-assisted attack against a U.S. retailer where hand-typed commands failed and the attacker switched to AI-generated scripts to achieve lateral movement in minutes. Defenders should expect compressed dwell time between initial access and domain compromise and plan detection and response against that timeline.
Check Point Gateway exploitation and ESXi wipers. Check Point Research confirmed that activity from 130.94.19[.]204 targeted Israeli organizations through vulnerable Mobile Access portals on Check Point Security Gateways roughly one week after a fix was available. BleepingComputer reported that successful compromises were followed by deployment of a VMware ESXi wiper. Separately, Check Point released emergency hotfixes for a high-severity zero-day granting root access and establishing persistence, exploited since late October 2026. Edge device exposure plus unpatched hypervisor hosts remains a complete kill chain.
N-able N-central and MSP targeting. N-able took N-central offline following disclosure of 33 flaws, including the 4EverAble2Shell chain enabling remote code execution and privilege escalation. Within hours of PoC release, at least 21 systems were compromised and used to deploy cryptocurrency miners, including 19 running the most recent security update. BleepingComputer also reported SCKit, a worm-like malware toolkit used in recent ScreenConnect attacks, designed to spread from infected customer devices to MSP technician workstations and then outward to other managed networks.
Supply chain and marketplace abuse of AI tooling. Cursor confirmed a security breach that enabled distribution of a malicious Claude Code extension through its marketplace after attackers stole credentials from AI infrastructure provider Fireworks AI. Microsoft published research on post-training supply chain attacks against coding agents, demonstrating how poisoned datasets and model checkpoints can trigger remote code execution. DarkReading reported that AI coding agents are leaking thousands of secrets to public GitHub repositories, with a Claude Code command inadvertently saving credential-bearing files.
GSocket C2 persistence on Linux. Microsoft detailed GSocket deployment on compromised Linux servers for encrypted remote access persistence after the initial vulnerability is patched. This is a common pattern where patching closes the entry point but leaves the implant in place; the writeup includes hunting guidance for GSocket artifacts.
GitHub account compromise at scale. A campaign dubbed wfx-loot used 4,837 compromised GitHub accounts to probe 61,764 repositories, open pull requests, and push secrets-stealing malware. The scale indicates automation for both account takeover and downstream repository manipulation.
Patch Tuesday volume. Microsoft's November Patch Tuesday included 1,328 vulnerabilities, 60 marked critical, 45 of which are remote code execution. Apache HTTP Server patched a critical HTTP/2 RCE exploitable without authentication. November updates also caused Windows activation failures with 0x80041008 error codes, which Microsoft has confirmed.
Defender Takeaways
- Treat edge devices as pre-compromised until patched and verified. The Check Point Gateway and N-able N-central cases both involved public exploitation within days of PoC release.
- Add hypervisor hunting to your scope. GhostFrame, Settra, and the ESXi wiper campaigns all operate inside ESXi where EDR does not reach. Collect ESXi logs centrally and alert on unexpected VIB installs and shell activity.
- Assume AI-driven intrusions will compress timelines. Tune detection and response SLAs against a three-hour initial-access-to-privilege window observed in Artex and Cairn cases.
- Review token theft paths alongside password theft. Luna Moth and device code phishing campaigns reported by Huntress show that MFA does not block session or access token exfiltration post-authentication.
- Audit AI coding agent configurations for inadvertent secret commits. The Claude Code and Cursor incidents both involved credentials and tokens leaving controlled environments through developer tooling.
- Track GSocket artifacts on Linux. Patching closes the door but does not evict the implant; validate that remediation includes host inspection.
- For MSP customers and providers, review ScreenConnect and N-central deployments for the SCKit and 4EverAble2Shell indicators published this week.
Sources
- ESET WeLiveSecurity, "Exchange for breakfast: Turla's Cereal tricks are not for kids," 2026-11-10. https://www.welivesecurity.com/en/eset-research/exchange-for-breakfast-turlas-cereal-tricks-are-not-for-kids/
- KrebsOnSecurity, "Who Runs the Ransomware Group Qilin?" 2026-11-11. https://krebsonsecurity.com/2026/11/who-runs-the-ransomware-group-qilin/
- Recorded Future, "DoppelCart Fraud Network Grows to More Than 250,000 Domains," 2026-11-10. https://www.recordedfuture.com/blog/doppelcart-fraud-network-grows-domains
- Recorded Future, "GhostFrame: A new ESXi Implant Linked to Settra Ransomware," 2026-11-10. https://www.recordedfuture.com/blog/ghostframe-esxi-implant-settra-ransomware
- Check Point Research, "PHP2Shell Exploitation on Check Point Gateway," 2026-11-12. https://research.checkpoint.com/2026/php2shell-exploitation-on-check-point-gateway/
- Check Point Research, "Not All That Glitters Is Gold: Tracking CopperDragon's Sparrow Framework Around the Globe," 2026-11-10. https://research.checkpoint.com/2026/not-all-that-glitters-is-gold-tracking-copperdragons-sparrow-framework-around-the-globe/
- Unit 42, "Attackers Use Autonomous AI to Target Semiconductor, Defense and Civil Society Sectors," 2026-11-12. https://unit42.paloaltonetworks.com/cairn-framework-ai-autonomous-attacks/
- Microsoft Threat Intel, "The global socket: how threat actors are weaponizing GSocket for stealthy C2," 2026-11-11. https://www.microsoft.com/en-us/security/blog/2026/11/11/global-socket-how-threat-actors-weaponizing-gsocket-stealthy-c2/
- Microsoft Threat Intel, "Trained to betray: How post-training supply chain attacks hijack coding agents," 2026-11-12. https://www.microsoft.com/en-us/security/blog/2026/11/12/trained-to-betray-how-post-training-supply-chain-attacks-hijack-coding-agents/
- Cisco Talos, "ARTEX and beyond: AI-enabled penetration testing platforms," 2026-11-13. https://blog.talosintelligence.com/artex-and-beyond/
- Sophos, "ARTEX: Attackers apply agentic AI," 2026-11-10. https://www.sophos.com/en-us/blog/artex-agentic-ai
- The Record, "FBI flash alert says Artex tool automating cyberattacks at 'machine speed'," 2026-11-13. https://therecord.media/artex-ai-tool-fbi-flash-alert
- BleepingComputer, "Check Point gateways hacked via PHP2Shell to deploy VMware ESXi wiper," 2026-11-13. https://www.bleepingcomputer.com/news/security/check-point-gateways-hacked-via-php2shell-to-deploy-vmware-esxi-wiper/
- BleepingComputer, "Check Point warns of zero-day flaw exploited for root access," 2026-11-13. https://www.bleepingcomputer.com/news/security/check-point-patches-zero-day-that-lets-hackers-get-root-access/
- BleepingComputer, "New N-able '4EverAble2Shell' zero-day exploited hours after PoC released," 2026-11-12. https://www.bleepingcomputer.com/news/security/new-n-able-4everable2shell-zero-day-exploited-hours-after-poc-released/
- BleepingComputer, "Cursor confirms security breach led to malicious Claude Code extension," 2026-11-14. https://www.bleepingcomputer.com/news/security/cursor-confirms-security-breach-led-to-malicious-claude-code-extension/
- BleepingComputer, "US disrupts North Korean campaign behind $400 million crypto theft," 2026-11-14. https://www.bleepingcomputer.com/news/security/us-disrupts-north-korean-campaign-behind-400-million-crypto-theft/
- BleepingComputer, "Hackers compromise 4,800 GitHub accounts to steal cloud credentials," 2026-11-12. https://www.bleepingcomputer.com/news/security/hackers-compromise-4-800-github-accounts-to-steal-cloud-credentials/
- BleepingComputer, "New worm-like 'SCKit' malware targets MSPs in ScreenConnect attacks," 2026-11-12. https://www.bleepingcomputer.com/news/security/new-worm-like-sckit-malware-targets-msps-in-screenconnect-attacks/
- BleepingComputer, "Apache HTTP Server patches critical HTTP/2 remote code execution flaw," 2026-11-12. https://www.bleepingcomputer.com/news/security/apache-http-server-patches-critical-http-2-remote-code-execution-flaw/
- DarkReading, "Extortion Group Luna Moth Flits Into Law Firms a Second Time," 2026-11-12. https://www.darkreading.com/data-privacy/extortion-luna-moth-law-firms-second-time
- DarkReading, "Inside an AI Attack Against a US Retailer," 2026-11-12. https://www.darkreading.com/cyberattacks-data-breaches/inside-ai-attack-us-retailer
- DarkReading, "AI Coding Agents Leak Thousands of Secrets to GitHub," 2026-11-10. https://www.darkreading.com/cyber-risk/ai-coding-agents-leak-secrets-github
- DarkReading, "N-able Takes N-central Platform Offline After Discovery of 33 Flaws," 2026-11-11. https://www.darkreading.com/endpoint-security/n-able-n-central-offline-33-flaws
- Huntress, "Crafty Microsoft Teams Booking Flow Disguises Device Code Phishing," 2026-11-12. https://www.huntress.com/blog/crafty-microsoft-teams-booking-flow-disguises-device-code-phishing
- Cisco Talos, "Microsoft Patch Tuesday for November 2026," 2026-11-11. https://blog.talosintelligence.com/microsoft-patch-tuesday-for-november-2026/
- CrowdStrike, "November 2026 Patch Tuesday: 1,252 CVEs Include One Zero-Day and 68 Critical Vulnerabilities," 2026-11-10. https://www.crowdstrike.com/en-us/blog/patch-tuesday-analysis-november-2026/