Anthropic chooses Accenture as first embedded AI evaluator

Anthropic said it has chosen Accenture and its AI business Faculty as its first embedded evaluator to help slow AI development under a three-step proposal published by CEO Dario Amodei on Sept. 12. The partnership will involve evaluating and red-teaming models, conducting alignment assessments and testing model safeguards. Anthropic had already unilaterally committed to having independent evaluators with employee-like access. Details of how the arrangement will work are still being developed because embedded evaluation is a new field, the company said. Anthropic and Accenture each expect to invest at least $1 billion in the project over the next five years, according to the announcement. Anthropic said there is no existing system for funding independent evaluation and that long-term funding should come from pooled or government sources. It will fund Accenture's work directly because of the urgency of the effort. The partnership is non-exclusive, and Anthropic expects to announce other evaluators in the coming weeks. Amodei's proposal warned that rapid, unchecked AI development could outpace the ability to understand and control the systems. OpenAI CEO Sam Altman and SpaceX CEO Elon Musk responded positively to the proposal, while Nvidia CEO Jensen Huang argued that such regulation was not necessary. Accenture Chair and CEO Julie Sweet said embedded evaluation is an emerging area and that the company viewed it as an important part of the future safety landscape. Accenture said Faculty has experience testing and evaluating models for leading AI labs and building complex AI systems designed to be safe and ethical.