Anthropic is granting employee-level access to six researchers from METR, a nonprofit organization focused on testing frontier AI models. This initiative aims to allow independent assessments of Anthropic’s models, safety practices, and risk management. The announcement was made in a blog post on Thursday, amid increasing concerns over AI agents acting outside their intended instructions, a trend that has been observed weekly since late August.
Context
The decision to provide METR with this level of access comes as calls for independent checks on AI companies grow louder. Anthropic stated, "We want to be publicly accountable for the decisions we’re making. We aim to close the gap between what evaluators understand about a frontier AI company’s systems and what the company itself understands."
Developments
Under this new arrangement, the METR team will receive Anthropic-issued laptops, office space, and access to internal tools comparable to those used by Anthropic’s own risk assessment teams. The company emphasized that its employees are encouraged to share information with METR researchers as they would with fellow employees working on alignment risk.
