Anthropic and Accenture Launch Embedded AI Safety Evaluator Team as Independence Debate Intensifies

Anthropic and Accenture said they will create an embedded AI safety evaluation team inside Anthropic, turning a fast-moving proposal for outside oversight of frontier AI labs into a concrete commercial partnership. The arrangement gives outside evaluators employee-like access to Anthropic's development process, while critics and evaluator groups warn that access alone does not guarantee independence.