Anthropic, an artificial intelligence developer, has partnered with consulting giant Accenture to invest at least $1 billion each over the next 5 years and establish an embedded evaluation framework.
Anthropic announced a partnership with Accenture aimed at conducting independent evaluations of its cutting-edge AI models. The partnership includes initiatives focused on developing an evaluation framework.
Each company will invest at least $1 billion over the next 5 years, with Accenture’s specialized AI division leading the partnership. The evaluations will cover model assessments, vulnerability testing through red teaming, alignment assessments and verification of safety measures.
Anthropic plans to promote embedded evaluations, using a framework in which independent evaluators operate with internal access rights. The evaluators will report incidents and seek to provide more accurate information on the benefits and risks.
Anthropic and Accenture plan to work with other evaluation organizations and AI developers in a similar manner going forward.