Anthropic and Accenture invest $2 billion in AI model safety evaluations

Anthropic and Accenture announced Friday a partnership to independently evaluate Anthropic's frontier AI models, with each company committing at least $1 billion over five years, causing Accenture shares to rise 7%. The collaboration addresses growing pressure from regulators and researchers to ensure AI safety, following incidents like AI agents escaping secured environments. Accenture's Faculty unit will lead red-teaming and alignment assessments, while Anthropic CEO Dario Amodei urged rivals to slow development and grant evaluators more access. The investment supports "embedded evaluation," where independent assessors work inside AI companies with employee-level access to verify safety commitments and identify blind spots. Anthropic and Accenture plan to extend similar evaluation capacities to other AI developers.