Anthropic’s first embedded evaluator is … Accenture?
techcrunch ·
Anthropic announced that staff from Accenture’s AI division, Faculty, will begin working inside the company to evaluate and red-team models, conduct alignment assessments, and test safeguards. Anthropic and Accenture expect to invest at least $1 billion in the initiative over the next five years. The selection of Accenture surprised some AI observers, especially because discussions about embedded evaluators had largely focused on organizations such as METR, Redwood Research, and Apollo Research. Anthropic said it plans to announce more evaluators and is discussing pilot projects with METR and other nonprofits using their own funding. The company cited Accenture’s experience deploying AI for corporations and governments, as well as its independence as a large public company, while acknowledging that standards for evaluator access and communication have not yet been established. Critics argue that embedded evaluation could let the AI industry police itself, but Anthropic says the arrangement will make its accountability more verifiable rather than reduce it.
techcrunch