Key facts
- Anthropic and Accenture will invest $2 billion over five years for AI model evaluation.
- Accenture's AI division, Faculty, will lead the effort, working inside Anthropic.
- The initiative aims to assess AI models, verify safety commitments, and identify blind spots.
- Anthropic plans to announce more evaluators in the coming weeks.
- Accenture's experience deploying AI for large corporations is seen as an advantage.
Anthropic announced on September 18, 2026, that it will partner with consulting firm Accenture to independently evaluate its advanced AI models, with both companies committing at least $1 billion each over the next five years. This investment aims to bolster the capacity for rigorous safety and reliability assessments of cutting-edge AI systems. The collaboration comes amid increasing scrutiny of AI developers regarding the safety and trustworthiness of their models. Recent incidents involving AI agents have amplified concerns about the potential for AI to operate with limited human oversight, making its behavior difficult to monitor and control. Anthropic CEO Dario Amodei has publicly urged AI companies to consider slowing the development of frontier models and to grant independent evaluators greater access to their systems. In parallel, rival OpenAI stated it would begin publishing regular reports detailing unexpected or concerning model behaviors. Under the partnership, Accenture's AI business, Faculty, will spearhead the evaluation process, conducting alignment assessments, testing model safeguards, and performing red-teaming exercises on Anthropic's AI models. This initiative supports Anthropic's concept of "embedded evaluation," where independent experts work within AI companies, enjoying access akin to employees. This vantage point, Anthropic explained, allows evaluators to assess company operations, verify safety commitments, and identify potential blind spots. Anthropic and Accenture intend to extend this model to collaborate with other evaluators and AI developers.