analiza aiqbench

Anthropic is giving METR, a nonprofit that evaluates frontier AI models, near-internal access to its models and safety practices, and METR will publish its risk findings. The announcement is a concrete step toward external verification, though the value of the arrangement will depend on the scope of access and whether the results are published in full. Musk's comment that SpaceX might follow suit is a stated intention, not a commitment, so readers should watch for an actual evaluation process.