Anthropic Partners with METR and SecureBio for AI Risk and Safety Assessments
AI firm Anthropic has engaged nonprofits METR and SecureBio to help assess risks and safety practices. This move is part of the company's commitment to embedding independent evaluators within its operations.
Artificial intelligence company Anthropic announced on Thursday that it is collaborating with nonprofits METR and SecureBio. The partnership aims to evaluate the risks and safety practices associated with Anthropic's AI models, fulfilling a commitment to integrate independent evaluators into the company's operations.
The move follows a proposal by Anthropic CEO Dario Amodei to slow the pace of AI model development, during which he pledged that third-party evaluators would help verify adherence to safety protocols. Days later, the company announced a partnership with Accenture, marking its first engagement with external safety assessors.
METR, known as Model Evaluation and Threat Research, previously operated in a less prominent niche within the AI industry. However, with mounting concerns about AI safety, the role of independent evaluators has become increasingly critical.
This initiative aligns with broader industry trends, including a voluntary accord proposed by President Donald Trump, which encourages AI companies to partner with external auditors. Anthropic emphasized its voluntary embedding of these evaluators, stating its belief that other leading AI firms should adopt similar practices.