Third-party evaluators found vulnerabilities in OpenAI monitoring system during test pilot
2026-10-22 · Tech
Third-party evaluators poked holes in OpenAI's safety monitoring system during a test run, revealing the company watching its own AI wasn't exactly airtight. Sam Altman's response? Promising evaluators 'employee-like access' — because nothing says 'we've got this under control' like needing outsiders to find the gaps you missed.