Anthropic is partnering with Accenture's Faculty division to conduct independent embedded evaluation of frontier AI models, including red-teaming and safety assessments. Both companies will invest at least $1 billion over five years, with embedded evaluators working inside Anthropic with employee-level access to assess model development, safety commitments, and identify risks while maintaining Anthropic's core responsibility for model safety.