Accenture Becomes Anthropic's First Embedded AI Evaluator

Anthropic has named Accenture as its first embedded safety evaluator, with staff from Accenture's AI division Faculty set to red-team models, assess alignment, and test safeguards inside the lab. Both companies plan to invest at least $1 billion over five years, sending Accenture shares up 8% after hours. The choice surprised many, as AI safety organizations like METR and Apollo Research were widely expected to fill such roles. Critics view the arrangement as industry self-policing, but Anthropic insists evaluators make its accountability more verifiable, not less. More evaluator partnerships are expected in coming weeks.