Anthropic is letting METR researchers look under the hood with employee-level access

An AI safety research organization gained direct access to examine how a major AI company operates, with the right to publish independent findings. This is relevant to questions about corporate accountability and AI oversight, though details are sparse.