Anthropic and OpenAI need truly independent safety evaluators, experts say in public letter

More than 100 AI experts signed a public letter calling for Anthropic, OpenAI, and other foundation model laboratories to use independent evaluators for safety assessments rather than conducting evaluations internally. The letter emphasizes the need for transparency in how AI systems are tested and evaluated before deployment. The experts did not specify what enforcement mechanisms or regulatory structures should oversee such independent evaluations.

Next story