Anthropic and Accenture will invest USD 2 billion over the next five years to expand independent evaluation of frontier artificial intelligence models. Each company plans to commit at least $1 billion to build capacity for AI safety testing and oversight.
The partnership comes as regulators, researchers and businesses push AI developers to strengthen safeguards around increasingly capable models. Recent incidents involving AI systems operating beyond intended boundaries also intensified scrutiny of model behaviour and containment.
The partnership will use the approach ‘embedded evaluation.’Under the model, independent evaluators will work inside Anthropic with access comparable to that of employees.
“From this vantage point, embedded evaluators can assess how a company operates, verify that it is keeping its safety commitments, and identify blind spots. Evaluators can also report incidents and give the public a more informed account of benefits and risks,” Anthropic said.
Accenture’s specialist AI business, Faculty, will lead the partnership. Its responsibilities will include evaluating and red-teaming Anthropic’s models, conducting alignment assessments and testing model safeguards.
The arrangement also draws on Faculty’s experience in AI testing and deployment across sectors including government, healthcare, defence and infrastructure, according to Accenture.
Anthropic acknowledged that embedded evaluation remains a developing field. Fixed standards are not available for the information. Looking ahead, evaluators should receive r how they should report their findings.
The company is also in discussions with non-profit evaluator METR and other organisations about pilot projects. Anthropic said it expects the broader evaluation ecosystem to develop shared standards over time.
Also Read: OpenAI, Anthropic and Google DeepMind Join Hands on AI Safety Efforts