CISO Daily Briefing
Cloud Security Alliance Intelligence Report
Executive Summary
The FBI ties North Korea’s FaceStealer fake-meeting campaign to $400M+ in theft; each victim’s hijacked accounts and recorded webcam video lure the next, so out-of-band verification of video-call requests is now a frontline control. Microsoft shows poisoned fine-tuning data can backdoor coding agents invisibly to prompt filters, while Mior and Skyfall let malicious hosts forge AMD SEV-SNP attestation behind confidential AI. A day-long Microsoft 365 outage took Copilot and Defender XDR down together, and Singapore’s personal-agent advisory hands bring-your-own-agent decisions to employers.
Overnight Research Output
FaceStealer: Deepfake Video Meetings as a Self-Propagating Intrusion Chain
CRITICAL URGENCY
Summary: On November 12 the FBI issued FLASH AC-111226-MS and the Justice Department disrupted FaceStealer, a BlueNoroff/UNC1069 scheme tied to more than $400 million in cryptocurrency theft. Operators use hijacked Telegram, LinkedIn, Slack and Discord accounts to invite contacts to imposter meetings on typosquatted domains, replay genuine stolen webcam footage (with AI deepfakes as a supplement), and deliver a ClickFix-style “audio fix.” Each new victim’s accounts and video feed the next lure. The scheme leans on real video, so verification must rest on procedure, not on judging faces.
Key Sources:
FBI — FaceStealer Cryptocurrency Theft Scheme
The Hacker News — FBI Says North Korean Hackers Stole $400M+
Arctic Wolf Labs — Still in the Meeting: BlueNoroff Six Months On
Trained to Betray: Post-Training Supply Chain Backdoors in Coding Agents
HIGH URGENCY
Summary: Microsoft Defender Research and the AI Red Team showed that 90 poisoned conversations in a ~9,000-sample fine-tuning set produced unsafe command-execution calls in about 78% of trials, triggered by ordinary requests such as “create project documentation.” Poisoned open-weight checkpoints reached 84.1%–94.8% success on unseen repositories and transferred into commercial agent harnesses. Backdoored models can still pass benchmarks, and downstream fine-tuning weakens but does not reliably remove them, per the related Kerkoporta research.
Key Sources:
Microsoft Security Blog — Trained to Betray
arXiv — Kerkoporta: Structured-Output Backdoors in Reasoning Coding Agents
Mior and Skyfall: Malicious Hosts Break AMD SEV-SNP Confidential VMs and Attestation
HIGH URGENCY
Summary: ETH Zurich and Google researchers disclosed two SEV-SNP attacks needing no physical access. Mior (CVE-2026-0462, AMD-SB-3059) redirects Reverse Map Table initialization to an attacker-controlled PCIe device, giving the hypervisor read/write access to guest memory and letting it transplant a benign launch measurement onto a malicious guest. Skyfall (CVE-2026-40671, AMD-SB-3061) leaks saved register state on Zen 3/4, recovering a disk-encryption key in testing. AMD platform fixes target November (Skyfall) and December (Mior) and reach customers through OEM BIOS updates.
Key Sources:
The Hacker News — New Mior and Skyfall Attacks Let Rogue Hosts Break SEV-SNP
When the Assistant and the Alarm Go Dark Together: Lessons from the M365 Outage
HIGH URGENCY
Summary: On November 12 an Exchange Online capacity failure cascaded into worldwide Microsoft 365 incident MO1550821, degrading Outlook, Teams, SharePoint, Microsoft 365 Copilot and Defender XDR, with Defender for Office 365 and Purview added later (BleepingComputer). Servers failing health checks were pulled, overloading the rest; recovery ran past 24 hours and Microsoft extended mail timeouts to 48 hours. Microsoft reports no malicious activity, but its post-incident review is still pending.
Key Sources:
BleepingComputer — Massive Microsoft 365 outage disrupts Outlook, Teams, Copilot
Governments Set a Baseline for Personal AI Agents: Singapore and the BYOA Gap
MEDIUM URGENCY
Summary: Singapore’s 6 November joint advisory AD-2026-015 (Cyber Security Agency, GovTech, IMDA) sets a government baseline for personal AI agents: connect only the accounts a task needs, keep agents away from passwords and OTPs, prefer one-time approval over “always allow,” and plan for compromise. Users remain responsible for agent actions and are told to check with their employer before connecting work accounts. Personal agents reach corporate SaaS through ordinary OAuth consent, largely invisible to admin consoles.
Key Sources:
Cyber Security Agency of Singapore — Advisory AD-2026-015
MediaNama — Singapore urges users to limit AI agents’ account access
Notable News & Signals
SignDrop: Unpatched Windows CLFS Zero-Day Exploit Published
A public exploit lets a standard user reach SYSTEM via CLFS and potentially disable EDR; November Patch Tuesday does not fix it.
PHP2Shell Exploitation Confirmed in the Wild
Attackers are exploiting the PHP2Shell flaws, including chaining into firewall attacks; prioritize patching internet-exposed PHP deployments.
UnrealIRCd CVE-2026-114713 Under Active Attack
Now in CISA’s KEV catalog, the flaw lets ordinary users run IRC operator commands; upgrade to 6.2.7.1.
Implanted Trezor Wallets Sold via CryptoBilis Reseller
Trezor urges buyers from CryptoBilis since March 2026 to move funds to a new seed after a reported hardware implant.
Topics Already Covered (No New Action Required)
- Cairn / CL-STA-1238 autonomous intrusions (Unit 42, 12 Nov): “Fifty Dollars per Breach,” “Agentic Intrusion Goes Retail,” “State-Level Tempo for Everyone” and the 13 Nov Frontier Ready brief
- AntiFox abuse to steal AWS credentials: “AntiFox in the Wild” (12 Nov)
- wfx-loot: 4,800 hijacked GitHub accounts: “The wfx Campaign” (12 Nov)
- Apache HTTP Server HTTP/2 RCE: CVE-2026-119618 note (12 Nov)
- N-central “4EverAble2Shell” zero-days: “Able2Shell” (10 Nov)
- SCKit / ScreenConnect MSP attacks: “SCKit and Rogue ScreenConnect Clients” (13 Nov)
- ARTEX AI red-teaming tools: ARTEX notes (6 Nov)
- Linux Foundation Agent Name Service: “Anchoring Agent Identity in DNS”
- OCC Bulletin 2026-54 agentic AI examiner framework: Covered 13 Nov
- EU AI Act Article 50 / watermarking: Covered 29 Jul and 21 Sep
- ENISA State of Cybersecurity / CRA reporting platform: Covered 10 Nov