Anthropic funds embedded AI evaluation at $1B scale

Anthropic and Accenture each plan to invest at least $1 billion over five years in independent evaluation of frontier AI, with Accenture’s Faculty unit evaluating and red-teaming models, conducting alignment assessments, and testing safeguards, according to Anthropic’s partnership announcement. The arrangement puts embedded evaluators inside Anthropic with access comparable to an employee’s, but Anthropic says standards for access, reporting, and funding are not settled; the partnership is non-exclusive. Builders should file this as an evaluation-governance signal, not a certification: use the archive’s model-evaluation governance analysis to separate independent evidence from vendor-controlled testing.