Anthropic said on September 18, 2026 that it is partnering with Accenture on independent evaluation of frontier AI, with each company expecting to invest at least $1 billion in building capacity in the area over the next five years.
The partnership will be led by Faculty, Accenture’s specialist AI business, and will include evaluating and red-teaming models, conducting alignment assessments, and testing model safeguards. Anthropic described the work as a step toward a commitment, made in a September 2026 essay by CEO Dario Amodei, to embed independent evaluators within the company.
Anthropic said Accenture helps businesses and governments deploy AI across many industries, and that the firm’s understanding of how enterprises use AI in practice informs its safety approach. That perspective, the company said, will be applied to evaluating its models.
How Embedded Evaluation Will Work
Embedded evaluation is new, and many of the details about how it will operate are still being worked out, according to the announcement. Where existing external evaluators work outside the companies they assess, embedded evaluators will work inside AI companies with access comparable to an employee’s.
That access, Anthropic said, allows evaluators to watch models take shape in training, follow the decisions that govern how models are built and deployed, and speak directly to employees. From that position, evaluators can assess how a company operates, verify that it is keeping its safety commitments, and identify blind spots. They can also report incidents and give the public a more informed account of benefits and risks, the company said.
Anthropic stated that independent embedded evaluators leave its accountability unchanged while making that accountability more verifiable, and that the safety of its models remains its responsibility.
Access Terms and Publication Rights
In the essay, titled “We Must Pace the Frontier”, Amodei proposed a three-step plan: embedded evaluators, coordination among frontier AI companies in democratic countries, and global coordination. He wrote that Anthropic was unilaterally committing to the first step and called on governments to require other frontier companies to match it. “We must slow the pace at which we improve the capabilities of AI models,” he wrote. “Progress will still seem fast, and we must make wise use of the time we gain.”
The essay says Anthropic intends to invite an embedded external review team equipped with desks in its offices, access badges, and company laptops, along with access to workspaces, tools, and permissions mostly comparable to what internal risk-assessment teams have. Exceptions would apply in cases such as where the law or contracts require it, or to protect customers’ and partners’ private information.
Under the contract approach described in the essay, external reviewers would hold the right to publish key findings about risk levels, incidents, practices, and the access they received or did not receive, without editorial control by Anthropic. The company would retain a narrow ability to redact security-sensitive, legally privileged, commercially sensitive, or third-party confidential information, but could not redact findings merely because they are unfavorable; reviewers could say publicly if a redaction removed something important to their conclusions. The essay points to precedent in the banking industry, where regulatory supervisors are sometimes embedded alongside employees.
Funding Questions and a Non-Exclusive Structure
Anthropic said there are as yet no standards for what information embedded evaluators should have access to, or how they should report what they find, and no settled system for funding independent evaluation. Long term, the company believes funding should come from pooled or government sources, as it called for in its Advanced AI Framework in June; as neither exists yet, it plans to work with different evaluators under different funding arrangements.
Anthropic will fund Accenture’s work directly. The company is also in dialogue with METR and other nonprofit evaluators to pilot elements of embedded evaluation using their own funding, and it said it believes frontier AI ultimately needs an ecosystem of evaluators operating with shared standards.
The partnership is non-exclusive. Anthropic said it will work with other evaluators it plans to announce in the coming weeks, that it expects frontier labs to work with several organizations at once, and that Accenture will work with other AI developers in similar capacities. The company said it will continue to train and release frontier models, is sharing its early efforts so people and other AI developers can see its process, expects its approach to evolve as the field matures, and will share more as work begins and additional evaluators come on.
Earlier Enterprise Partnership
The evaluation work extends a broader relationship the two companies announced on December 9, 2025: a multi-year partnership that formed the Accenture Anthropic Business Group. Under that agreement, approximately 30,000 Accenture professionals are to receive training on Claude, Claude Code is being made available to tens of thousands of Accenture developers, and the companies launched a joint offering designed to help CIOs measure value and drive AI adoption across their engineering organizations. The December 2025 announcement also included initial industry solutions for regulated sectors, among them financial services, life sciences, healthcare, and the public sector.
Credit: Source link
