Partnering with Accenture on embedded evaluation
Accenture will work inside Anthropic to evaluate frontier models, with Anthropic directly funding the effort.
The work will be led by Faculty, Accenture’s specialist AI business, and includes red-teaming, alignment assessments, and safeguard testing. Anthropic says embedded evaluators will get employee-like access to models in training, internal decisions, and staff. The company says the setup is meant to make its safety commitments more verifiable, while responsibility for model safety remains with Anthropic. It also says the partnership is non-exclusive and that other evaluators will be announced in the coming weeks. Anthropic's note
The work will be led by Faculty, Accenture’s specialist AI business, and includes red-teaming, alignment assessments, and safeguard testing. Anthropic says embedded evaluators will get employee-like access to models in training, internal decisions, and staff. The company says the setup is meant to make its safety commitments more verifiable, while responsibility for model safety remains with Anthropic. It also says the partnership is non-exclusive and that other evaluators will be announced in the coming weeks. Anthropic's note
score 7