Anthropic Names Accenture Its First Embedded Evaluator, With Staff-Level Access to Model Development

Accenture and Anthropic announced on September 18, 2026 that they are partnering to establish a team of embedded evaluators who will work alongside Anthropic's internal teams and safety partners to evaluate and red-team models, conduct alignment assessments and test model safeguards.

Evaluators With Employee-Level Access

The team is led by Faculty, Accenture's specialist AI business, and will have internal access comparable to Anthropic staff. That access is intended to let evaluators observe models during training, examine development and deployment decisions, and talk directly with Anthropic employees — a much deeper level of visibility than the pre-release testing most outside evaluators get today.

At Least $1 Billion From Each Side

Both companies expect to invest at least $1 billion over five years in AI safety, and Anthropic will fund Accenture's evaluation work directly. CNBC reported that the program is designed to help put into practice the slowdown proposal put forward by Anthropic CEO Dario Amodei.

Non-Exclusive, With More Evaluators to Come

The arrangement is non-exclusive. Anthropic said it expects to announce other evaluators in the coming weeks, which suggests embedded evaluation is meant to become a standing layer of oversight rather than a one-off engagement with a single firm.

Why Enterprises Should Pay Attention

For companies buying AI, independent evaluation is becoming part of how frontier models are developed rather than something done only after release. Expect procurement and risk teams to start asking every model vendor the same question: who outside the company has looked inside, and what did they find?