Anthropic and OpenAI Propose Embedding Independent Safety Evaluators Inside AI Companies

A proposal to place independent safety evaluators within leading AI labs could strengthen oversight of frontier model development.

Auto-scored: A real but partial governance step, as it is only a proposal without binding force or implementation.