Anthropic CEO Dario Amodei Calls for Slowing AI Development, Opens Models to External Evaluators

A three-step plan: first, granting external evaluators like METR broad model access; second, unifying industry efforts; third, government regulation to slow training and development.

Anthropic argues the current pace outruns companies' ability to build safety safeguards and regulators' ability to evaluate models, creating systemic risks.

It may push countries like the UAE and Saudi Arabia to accelerate their AI regulatory frameworks and open the door to collaboration with international evaluators like METR.

No, it is a voluntary unilateral step by Anthropic and would require industry coordination and government legislation to become binding.