Anthropic has named Accenture as an independent evaluator of its AI models for safety, with each company investing at least $1bn over a five-year partnership. The collaboration will be led by Faculty, a UK AI company acquired by Accenture for $1bn in January, under Marc Warner, now Accenture's chief technology officer and former UK AI Council member. Warner emphasized AI must be "safe by design, not safe by accident." Accenture shares rose about 7 per cent in after-hours trading.

The agreement follows Anthropic CEO Dario Amodei's call for a coordinated slowdown in AI research and third-party model evaluation, supported by Sam Altman of Open AI and Elon Musk of x AI. Accenture's team will evaluate Anthropic's Claude models as a Red Team, testing safeguards and assessing alignment research. Anthropic will fund Accenture's work, though critics question evaluator independence from leading AI labs. Anthropic aims for long-term funding from "pooled or government sources" and is also engaging non-profit evaluators like MET R.

Preparing for an IPO potentially valuing it over $2tn, Anthropic stated the partnership is non-exclusive and hopes other AI developers join. However, Meta and Amazon have not fully endorsed Amodei's proposals, favoring self-regulation. Accenture brings experience from trust and safety work for Facebook and You Tube.