Updated

Anthropic CEO Dario Amodei published an essay arguing it is time to 'pace the frontier,' proposing a three-step plan to slow AI training and development so companies can build safeguards and regulators can properly evaluate models. He said Anthropic will give third-party evaluators such as METR expanded access to its models to verify adherence to safety commitments. Amodei's stance appears to align with recent comments from OpenAI's Sam Altman about the risks of uncontrolled AI advancement, signaling a notable moment of convergence among leading AI labs on the need for caution even as competitive pressure to advance rapidly continues.