Anthropic Gives METR Employee-Level Access to Review AI Risks

Six researchers from nonprofit METR will work inside Anthropic with access similar to its risk teams, although customer data, model architecture and some other information remain off limits.