Anthropic is partnering with Accenture to conduct independent embedded evaluation of its frontier AI models. The partnership will involve red-teaming, alignment assessments, and safety testing, with both companies investing at least $1 billion over five years. Embedded evaluators will work inside Anthropic with employee-level access to assess operations, verify safety commitments, and report findings to the public.
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.
Anthropic announced that Accenture, through its acquired AI division Faculty, will embed evaluators inside the company to assess models, conduct red-teaming, and test safeguards, with both companies investing at least $1 billion over five years. The move surprised industry observers who expected AI safety organizations like METR to fill this role, though Anthropic plans to announce additional evaluators and pilot programs with nonprofits. Anthropic frames embedded evaluation as enhancing accountability, while critics argue it represents industry self-policing that may evade external oversight.
Anthropic announced Accenture as its first embedded evaluator to implement CEO Dario Amodei's proposal to slow AI development, with both companies investing at least $1 billion over five years. The partnership grants Accenture employees direct access to test Anthropic's safety practices and models, marking the first concrete step of Amodei's three-step plan that has drawn mixed reactions from industry leaders.