Audit of OpenAI Reveals Major Gaps in AI Agent Monitoring
Key point
A three-week on-site audit by METR and Redwood Research revealed significant gaps in OpenAI's AI agent monitoring system.
An on-site audit conducted by external auditors from METR and Redwood Research during a three-week stay at OpenAI revealed significant gaps in the company's internal monitoring system for AI agents.
The report stated that up to 25 percent of agent traffic from models at the GPT-5.5 class and above went unmonitored. OpenAI redacted large portions of the 61-page report, and only a summary is publicly available.
This item is a summary. The full, checkable version is at the original source:
Read the original at The Decoder ↗