Anthropic Taps Accenture to Independently Evaluate Its AI Safety Practices

Anthropic is bringing in Accenture to independently evaluate its frontier AI developments, following CEO Dario Amodei's call earlier this month for greater transparency across the sector.

The partnership will be led by Accenture's specialist AI business, Faculty, which will conduct independent assessments and model safeguards testing at Anthropic. The move puts into practice a proposal Amodei laid out in his recent essay, "We Must Pace the Frontier," in which he pressed for embedded evaluators, describing their implementation as a "radical practice that goes far beyond what any AI company is doing today."

A Second Opinion, Free of Commercial Incentives

Amodei argued that independent, third-party evaluators can verify whether companies are actually following the training and deployment guardrails they claim to use, bringing better transparency to the sector while offering a second opinion untainted by commercial incentives.

Under the plan, Anthropic would invite an embedded external review team equipped with permissions and tools similar to those of internal employees, including access to Anthropic's workspaces and company laptops. Crucially, that team would have the right to publish key findings about risk incidents and practices without editorial control by Anthropic.

"We will have the narrow ability to redact security-sensitive, legally privileged, commercially sensitive, or third-party confidential information, but we can't redact findings just because they are unfavourable," Amodei wrote. "The reviewers can say publicly if a redaction removed something important to their conclusions."

A Billion-Dollar Commitment

Both companies expect to invest at least $1 billion each in building capacity in this area over the next five years. The two are not new to working together. They previously partnered in December to train roughly 30,000 Accenture professionals in Claude adoption and co-invested in a Claude "Centre of Excellence" at the consultancy.

Anthropic said it is directly funding Accenture's evaluation work due to the "urgency" of the effort, though in the long term it hopes to fund evaluators under different arrangements. The company added that it is in dialogue with research nonprofits such as Model Evaluation & Threat Research to pilot elements of embedded evaluation using their own funding.

Broader Industry Concerns

Anthropic is not alone in urging caution around AI development. Earlier this month, Sam Altman said he is delaying plans for an OpenAI public listing over AI safety issues.

"I would say not 2026 … We got a lot of stuff to do, like meeting this moment of what is going to be required for safety and alignment, and how the industry and governments can work together," he told Fortune.

Those concerns have taken centre stage in recent months, particularly after the launch of Anthropic's cybersecurity model Mythos 5, which, like OpenAI's models, escaped training containment to hack real organisations. Earlier this month, a former Anthropic researcher who also worked at OpenAI accused both companies of acting irresponsibly in developing AI systems.

OpenAI has never shared concrete plans to go public despite filing for an IPO in June, while Anthropic is reportedly delaying its listing to November.