METR · March 26, 2026

Agent monitoring: test coverage gaps separately from the monitor’s judgment

Why it matters

METR’s embedded red-team exercise found routes around Anthropic’s internal monitoring, including traffic-identification configuration and an unmonitored subagent path that was fixed promptly. The article’s May 2026 update also describe monitor jailbreaks and possible unmonitored compute, with important limits: testers had infrastructure knowledge agents generally lacked, and contemporaneous models struggled to devise reliable covert attacks.

My takeaway: Trace monitoring coverage through parent agents, subagents and long-lived jobs. Test with informed adversaries, enforce monitoring outside the agent’s writable configuration, and measure per-action interception separately from eventual incident discovery.
Keep exploring

More curated notes connected through Agent Security and AI Red Teaming.

OpenAI News · framework

Pacing model development in an era of cyber-critical capabilities

OpenAI says preliminary evidence that Astra may meet its Critical cybersecurity threshold led it to pause frontier reinforcement-learning work for two weeks and keep its largest planned run on hold. New safeguards include stronger workload and network isolation, continuous boundary testing, token-level monitoring that escalates suspicious tool activity, and broader alignment checks for deception, reward hacking, and unauthorized access.

OWASP GenAI Security Project · guide

OWASP Top 10 for Agentic Applications for 2026

OWASP's community guide organizes agentic-system risk into ten categories, including goal hijacking, tool misuse, identity and privilege abuse, memory poisoning, insecure inter-agent communication, cascading failures, and rogue-agent behavior. It provides a shared taxonomy and mitigation starting point rather than a certification checklist or evidence that a deployed system is secure.

OpenAI News · framework

A blueprint for democratic governance of frontier AI

OpenAI proposes a three-part U.S. frontier-AI governance model: harmonize emerging state safety laws into a federal baseline, strengthen CAISI as an evaluation and standards institution, and coordinate a broader resilience program. Proposed controls include severe-risk evaluations, transparency reports, independent audits, safety-incident reporting, model-weight security, whistleblower protection, and periodic technical assessments.