independent review

Anthropic is allowing researchers from Model Evaluation and Threat Research to inspect its internal systems using employee-level access.

Anthropic gives six METR researchers employee-level access to assess AI safety and risk, with independent publ…