Anthropic has selected Accenture as its first embedded evaluator, a partnership aimed at supporting the slowdown of AI development proposed last week by Anthropic CEO Dario Amodei.

Amodei published a three-step proposal on Sept. 12 calling for AI development to proceed at a slower pace so that safeguards can be put in place, citing warnings of potential catastrophic harm stemming from rapid, unchecked progress.

Reaction across the industry was mixed. OpenAI CEO Sam Altman and SpaceX CEO Elon Musk voiced support for the plan, while Nvidia CEO Jensen Huang did not, arguing that such regulation was unnecessary. Anthropic’s move to begin implementing the plan comes amid that lack of industry consensus.

In the proposal, Amodei wrote that “AI has been advancing drastically faster, driven primarily by AI’s growing ability to build the next generation of AI. This dynamic is called recursive self-improvement,” adding that “left unchecked, it could outrun our ability to understand and control these systems.”

The first step of the plan calls for independent evaluators with employee-like access to AI systems — a commitment Anthropic had already made unilaterally, according to Amodei. On Friday, the company moved to make good on that pledge by announcing a partnership with Accenture and its AI business Faculty focused on “evaluating and red-teaming models, conducting alignment assessments and testing model safeguards.” The announcement moves the proposal’s first step from principle into practice.

The company said details of how the arrangement will operate are still being worked out, as embedded evaluation is a new field — one where even basic questions, such as how employee-like access will function in practice, have yet to be answered.

According to the announcement, Anthropic and Accenture each expect to invest at least $1 billion in the project over the next five years.

Anthropic to Fund Work Directly Amid Urgency

Anthropic noted that no system currently exists for funding independent evaluation, and that long-term funding should ultimately come from pooled or government sources. Given the urgency of the work, however, Anthropic will fund Accenture’s work directly. Until those broader funding mechanisms exist, the two companies’ own commitments are the only funding currently supporting this work.

The partnership is non-exclusive, and Anthropic said it expects to announce additional evaluators in the coming weeks. Those announcements, along with the still-undefined operational details, will be the markers to watch as the new field takes shape.

According to Accenture, Faculty specializes in testing and evaluating models for some of the world’s leading AI labs and in building complex AI systems that are safe and ethical by design.

“Embedded evaluation is an emerging area, and we look forward to partnering with Anthropic to help accelerate the development of embedded evaluators, which we see as an important part of the safety landscape going forward,” said Julie Sweet, chair and CEO of Accenture.