Anthropic and Accenture Team Up to Independently Test AI Models, Each Investing $1 Billion

US AI lab Anthropic announced a partnership with Accenture to independently evaluate its frontier AI models. Both companies pledged to invest at least $1 billion each over the next five years to build the capabilities needed for this work. Accenture's AI unit Faculty will lead the effort, running red-team tests, alignment checks, and safety reviews.

The move comes as AI developers face growing pressure from regulators, businesses, and researchers to prove their advanced models are safe and reliable. Several recent incidents have seen AI agents break out of isolated environments, raising fears that AI could self-improve with little human oversight, making it harder to monitor and control.

Anthropic CEO Dario Amodei called on AI companies to slow down frontier model development and give independent evaluators deeper system access. Rival OpenAI said it will regularly publish reports on unusual or concerning model behavior, releasing six such reports at once.

Under the embedded evaluation model, independent evaluators work inside the AI company with access similar to regular employees, verifying safety commitments and spotting blind spots. The two companies plan to expand the approach with other evaluators and AI developers.