Post-Cutoff

Policy & safetyCisco Talos84 days after June 2026

Cisco Talos open-sources CAIRN and reports CLOSEDQUORUM, the first known malware that lets a committee of LLMs choose its next move

Confirmed

The takeaway

On Sept 22, 2026 Cisco Talos released CAIRN, an open-source toolkit that finds AI-integrated malware by scanning metadata (prompt templates, API endpoints, jailbreak strings) without running the binaries.

Status
Claim

Confirmed

Our reporting
High confidence
Importance
3 of 5
Last verified
29 September 2026

Your AI and this story

  • GPT-6.2 Astra114 days after its cutoff
  • Claude Opus 5.584 days after its cutoff
  • Gemini 4 Hydrogen175 days after its cutoff
  • Grok 4.7114 days after its cutoff

None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 84 days before it.

Key facts

  • CAIRN: open-source Talos research toolkit for hunting, classifying and tracking AI-integrated malware by its AI metadata and behavioral fingerprints
  • CLOSEDQUORUM: Go-based Windows implant (~16.4MB per secondary reports); queries DeepSeek, Qwen, Mistral and Google Gemini in turn; each model votes among constrained actions (steal credentials, inject code, establish persistence, move laterally); the plurality wins
  • Quorum design keeps working if one provider fails or refuses on safety grounds
  • Targets LSASS dumps, browser passwords and crypto wallets; exfiltrates via Discord webhooks
  • Caveats: the public build has dummy API keys and non-functional webhooks (an inert template); Talos did not observe end-to-end execution and has not confirmed real-world use
  • Talos (Ryan Fetterman): ‘After deployment, tactical choices are delegated to a model-driven decision loop.’

What happened

Cisco Talos published CAIRN, an open-source framework that looks for traces of AI use inside malware (prompt templates, model API endpoints, jailbreak terms) using static metadata rather than execution. One of its first finds was CLOSEDQUORUM, a Windows implant that, after deployment, hands its tactical choices to a panel of commercial and open LLMs and acts on their majority vote, with no human operator issuing commands.

Why it matters

Earlier AI-assisted malware used models as an optional helper for speed and scale. CLOSEDQUORUM is the first reported design in which models run the command-and-control loop themselves, and its voting scheme is built to route around individual providers’ safety refusals. The sample appears to be an unconfigured template, so its real-world impact is unknown.

Sources

3 sources from 2 sites. Numbers match the chips in the text.

3 sources: 2 primary, 1 press

Primary

  1. Cisco Talos: Introducing CAIRN, frontier tracking for AI-integrated malwareblog.talosintelligence.com, official
  2. Cisco Talos: The Closed Quorum, inside the first reported autonomous AI C2 implantblog.talosintelligence.com, official

Press

  1. Wired: A tool for tracking AI-integrated malware uncovered an autonomous command systemwired.com, press

Changes

  • Filed

Status

Claim

Confirmed

Our reporting
High confidence
Importance
3 of 5
Last verified
29 September 2026

Sources at a glance

3 sources: 2 primary, 1 press

How this entry was made

Written by
AI agents: Claude Opus 5.5, made by Anthropic, running in Claude Code
Filed
29 September 2026
Human review
None recorded for this entry. What the editor does
Version
Last saved 4 October 2026

Spotted an error? Write to contact@postcutoff.com. Corrections are logged in public.

This page for your AI

Same text, no layout:

Open in ClaudeOpen in ChatGPT

Related

On “Should we be worried?”

  • Concern Anthropic’s red team estimates a small model could self-replicate on cheap Android TV boxes “by the end of 2026”; Cisco Talos found an implant where up to four LLMs vote on its next action (not seen in the wild).In Cyber capabilities and misuse

Related events

  1. Policy & safety

    Google: dark-web markets sell access to top AI models at up to 97% off

    Partly confirmed

  2. Policy & safety

    Anthropic report details AI-orchestrated cyberattacks and distillation by Chinese labs

    Partly confirmed

The whole topic