Anthropic has named Accenture as its first embedded evaluator, marking a significant move to implement CEO Dario Amodei’s recently announced plan to slow down AI development. This partnership involves integrating Accenture employees into Anthropic’s teams to rigorously test and verify the safety of its AI models. Both companies have committed to investing at least $1 billion over five years to strengthen capabilities in AI safety, with Anthropic directly funding Accenture’s initial work due to the critical urgency of the initiative.

Amodei’s three-step slowdown proposal, published in early September 2026, calls for external evaluators to have employee-level access within AI firms to monitor and report safety concerns. While Anthropic is taking the lead with this embedded evaluator approach, it invites other AI companies to follow suit in adopting similar transparency and monitoring practices. The company has emphasized that working with external evaluators like Accenture will not lessen its ultimate responsibility for the safety and alignment of its AI models.

In response to growing fears about AI’s potential catastrophic risks, Anthropic and OpenAI have faced heightened scrutiny from researchers and regulators worldwide. Amodei’s plan has garnered mixed reactions within the tech community, receiving endorsement from figures such as OpenAI CEO Sam Altman and Elon Musk, while others like Nvidia CEO Jensen Huang remain skeptical about the need for additional regulation at this stage. Meanwhile, Anthropic is preparing for a major IPO, intensifying attention on how it manages AI development risks.

Anthropic clarified that its collaboration with Accenture’s Faculty division is non-exclusive, with ongoing talks involving other third-party evaluators like the research nonprofit METR. The company plans to evolve its evaluation approach over time and publicly share progress to encourage industry-wide best practices. This pioneering step illustrates Anthropic’s commitment to creating a safer AI future by embedding independent safety oversight directly within its operational framework.