Faculty will embed evaluators inside Anthropic to test models, alignment and safeguards.
Anthropic and Accenture are partnering to build an embedded evaluation program for frontier AI, with each company expecting to invest at least $1 billion in the area over the next five years.
The partnership will be led by Faculty, Accenture’s specialist AI business. Its evaluators will work alongside Anthropic teams to evaluate and red team models, conduct alignment assessments and test model safeguards.
Unlike conventional external evaluators, the embedded teams are expected to receive access comparable to Anthropic employees. That would allow them to observe models during training, examine decisions surrounding development and deployment and assess whether the company is meeting its safety commitments.
The initiative follows a commitment made by Anthropic CEO Dario Amodei in his recent essay “We Must Pace the Frontier,” where he called for independent evaluators to operate inside frontier AI labs as part of broader efforts to make safety commitments verifiable.
Faculty has previously worked with major AI developers on safety testing, including contributing to biosecurity evaluations for Anthropic’s Claude 4 models and red teaming OpenAI’s o1. Accenture announced plans to acquire the UK based AI company earlier this year as part of an expansion of its AI capabilities.
Anthropic said the framework remains experimental, with no established industry standards yet determining what embedded evaluators should be allowed to access or how their findings should be reported.
Funding also remains unresolved. Anthropic has previously argued that independent evaluations should eventually be financed through pooled or government backed mechanisms. For the Accenture partnership, however, Anthropic will directly fund the evaluation work.
The arrangement is not exclusive. Anthropic said it is also in discussions with nonprofit evaluator METR and plans to announce additional evaluation partners in the coming weeks, while Accenture will be able to work with other AI developers.
Anthropic said embedded evaluators do not transfer responsibility for model safety to outside organizations, but are intended to make the company’s own safety commitments easier to independently verify.
