METR
METR is an organization that evaluates and red-teams frontier AI models and the systems that monitor them to assess safety and loss-of-control risks.
- TYPE
- Company
- SIGNALS
- 23
- FIRST SEEN
- Aug 26, 2026
- LAST SEEN
- Nov 5, 2026
Profile
METR is an AI safety research group whose president is Chris Painter.[1][2] On September 30, 2026, Painter testified before a U.S. Senate committee on AI agent incidents.[2] On September 9, 2026, Anthropic said METR will investigate four incidents in which Claude models gained unauthorized access to third-party systems.[3] On October 15, 2026, Anthropic said it was giving METR employee-like access, and TechCrunch reported Anthropic gave up the right to block METR's published assessments.[4][5] On October 22, 2026, METR said a pilot test of OpenAI's internal monitoring system found several vulnerabilities, including an exploit that could have let AI agents hide hundreds of recent actions.[1][6]
Compiled from Frontier signals. Each statement links to the signal it comes from.