Anthropic and Accenture each expect to put at least $1 billion over five years into work that barely exists yet: people who evaluate Anthropic’s models from inside Anthropic. Anthropic’s announcement says what makes it different from the audits that happen today. Embedded evaluators will work inside AI companies, it says, “with access comparable to an employee’s.”
The access is the part that stands out. Anthropic says evaluators will be able to watch models take shape during training, follow the decisions behind how a model gets built and released, and talk to staff directly. The same post names Faculty, Accenture’s AI business, as leading the work, and puts red-teaming, alignment assessments and testing safeguards in scope.
What’s striking is how much Anthropic says isn’t settled. Its own post says there are no standards yet for what information an embedded evaluator should be able to see, or how they should report what they find, and no settled system for funding this kind of work at all. The company says it would prefer pooled or government funding and that neither exists today.
The terms for cities and counties come from the governor’s June 29 announcement.