Anthropic and Accenture reportedly plan at least $2 billion for AI safety evaluation

A post citing Reuters’ Japanese edition says Anthropic and Accenture plan to invest at least $2 billion combined over five years in model evaluation. Another post, describing a September 18 announcement, identifies Faculty as Anthropic’s first “embedded evaluator,” with employee-like access to observe training, follow deployment decisions, test models for weaknesses and report incidents. It puts each company’s commitment at at least $1 billion and argues that Anthropic’s direct funding makes evaluator independence central to the arrangement.