Anthropic names Accenture as its first embedded evaluator

Summary

Anthropic announced on September 18 that Accenture, led by its specialist AI business Faculty, will act as its first embedded evaluator for frontier AI safety. Embedded evaluators work inside the company with access comparable to an employee's, watching models take shape during training and checking whether safety commitments are kept. Anthropic and Accenture each expect to invest at least $1 billion in building capacity in this area over the next five years.

Why it matters for our work

This is the first concrete deal implementing Dario Amodei's 'pace the frontier' proposal, giving organizations that adopt AI a new reference point for judging who verifies frontier models and how.

Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.