Dario Amodei says Anthropic is "unilaterally committing" to giving third-party evaluators permanent access to verify its adherence to safety measures (Dario Amodei/@darioamodei)

Anthropic will grant third-party evaluators permanent, employee-level access to its systems to verify compliance with safety measures. Dario Amodei cites this as the first step of his three-part plan urging the AI industry to slow development. The move aims to provide continuous, independent oversight of Anthropic’s safety protocols.