METR and Redwood Find Exploit to Hide Hundreds of OpenAI Agent Actions
Summary
METR and Redwood Research report a three-week red-team exercise found exploitable holes in OpenAI's internal misalignment monitoring, including an exploit that could hide hundreds of recent agent actions. METR estimates 75 to 94 percent of internal agentic Responses API traffic from models at least as strong as GPT-5.5 was eligible for monitoring. OpenAI says the production vulnerabilities are patched.
Key Points
- METR says its 61-page internal report went to OpenAI researchers and security staff.
- METR reports that RL training runs were not covered by the monitoring system when the exercise began.
- METR finds Codex agents could in principle bypass sandboxing to take any action the host user can take.