Anthropic and Accenture commit $1 billion each to model evaluation
Accenture's Faculty unit will get access comparable to an employee's. The arrangement is not exclusive, and Anthropic says talks with non-profits continue.
Anthropic said on Thursday it will work with Accenture on independent evaluation of its frontier models, with each company expecting to invest at least $1 billion in the effort over five years. The work will be led by Faculty, Accenture's specialist AI unit.
The scope covers evaluation and red-teaming of models, alignment assessments and tests of model safeguards. Anthropic said Faculty will have access comparable to that of an employee, which is the part that distinguishes this from an external audit conducted at arm's length.
The company framed it as a step toward a commitment made in an essay by its chief executive, Dario Amodei, to place independent evaluators inside the company. The arrangement is non-exclusive, and Anthropic said it is in discussions with METR and other non-profits about piloting similar access using their own funding.
The choice of partner drew immediate comment. TechCrunch questioned whether a large consulting firm is the natural candidate for independent scrutiny of a frontier laboratory, given that Accenture also sells AI services commercially.
The commitment is Anthropic's own announcement and the figures are the companies' own. No external body has verified how the access will work in practice, and neither company has said what Faculty would be able to publish.