Anthropic has said that staff from Accenture will begin working inside the company to scrutinise its models and staff, in the first embedded evaluator arrangement to emerge from chief executive Dario Amodei's plans to place third-party safety evaluators inside AI labs.
In a blog post, Anthropic said Faculty, a company Accenture acquired in January to act as its AI division, will begin "evaluating and red-teaming models, conducting alignment assessments, and testing model safeguards." Both companies expect to invest at least $1 billion in the project over the next five years.
The choice of Accenture surprised many AI watchers, and the markets, where the consulting company's shares rose 8% after hours. Discussion of embedded evaluators had previously focused on AI safety research organisations such as METR, Redwood Research and Apollo Research.
Anthropic pointed to Accenture's practical experience deploying AI for large corporations and government agencies as a key advantage, and noted that as a large public company predating the AI boom it is more functionally independent of Anthropic and the ecosystem around the lab. The lab said no standards yet exist for evaluators' access or communications and that it expects its approach to evolve over time.
Anthropic said more evaluators will be announced in the weeks ahead and that it is in conversation with METR and other non-profit organisations about how to "pilot elements of embedded evaluation using their own funding." Some critics see the self-policing scheme as a way to evade accountability for AI model misbehaviour. Anthropic insists the evaluators "do not reduce our accountability, but help to make it more verifiable," adding that "the safety of our models remains our responsibility."