Anthropic and Accenture Announce a USD 2 Billion Independent AI Model Evaluation Partnership
Anthropic and Accenture announced a partnership on 18 September 2026 for independent evaluation of frontier AI models, each committing at least USD 1 billion over five years for a combined USD 2 billion. Accenture's specialist AI unit Faculty will lead the work, evaluating Anthropic's models through red-teaming, alignment assessments and safeguard testing. The deal backs embedded evaluation, where outside evaluators work inside an AI company with employee-level access to verify safety commitments. Accenture's shares rose 7% in extended trading, and Anthropic CEO Dario Amodei urged slower frontier development with wider access for independent evaluators.
Dario Amodei is the Chief Executive Officer of Anthropic | Red-teaming means deliberately probing a system for failures and unsafe outputs | Alignment refers to how closely a model's behaviour matches human intent and safety requirements