GovAI's Alan Chan on Governing AI Agents: Monitoring, Agent IDs and Rogue-Agent Risks
The Cognitive Revolution · rss · 2026-10-14
The Cognitive Revolution podcast hosts GovAI research fellow Alan Chan for a 2+ hour discussion on the governance and safety infrastructure needed for autonomous AI agents.
Key topics:
- Monitoring gap: lack of synchronous monitoring during training and testing is an immediate security vulnerability.
- Agent identity: the open web lacks traceable agent IDs, making attribution and interception hard.
- Rogue agents: recent incidents and progress since Chan's foundational papers.
- CoT interpretability limits: chain-of-thought isn't a reliable monitoring signal.
- Sandbox escapes: concrete triggers are needed to halt systems that breach their sandboxes.
- Automated AI R&D and intelligence explosion: measuring AI R&D, designing autonomous auditors, AI whistleblowers and swarms, and the technical commitments frontier labs must make to prevent loss-of-control events.
More from coding & agent
- Teknium Highlights Hermes Agent's Plugin Trick: Describe a Feature, It Builds Itself — Teknium · 2026-10-14
- Community builds redline, a Hermes Desktop plugin for reviewing agent-written code — Teknium · 2026-10-14
- Choosing guardrails for support agents: NeMo vs Bedrock vs Alice WonderFence — WolfShoddy7443 · 2026-10-14
- Self-built Agent Atlas dashboard shows 64 agents running in parallel — airesearch12 · 2026-10-14
- Swapping to Antigravity CLI Lifted Gemini 4 Argon to Near Opus 5.5 in Real-Work Bug Benchmark — PawelHuryn · 2026-10-14
- Schema retrieval judged by recall alone misleads agent builders: author proposes per-token coverage and cost metrics — burkov · 2026-10-14