Anthropic chief executive Dario Amodei published a proposal to slow AI development. He calls it "We Must Pace the Frontier". It rests on three pillars. First, outside evaluators with employee-level access to monitor risk. Second, safety standards agreed between democratic states. Third, coordination with authoritarian governments, including China, on security. Sam Altman of OpenAI said he would also embed independent evaluators. Demis Hassabis of Google DeepMind and Elon Musk of xAI backed the proposal too. Critics say the firms want to pick their own evaluators and head off real rules. Former officials David Sacks and Alvaro Bedoya called it an attempt at regulatory capture.

• $24m spent on one House campaign
• 3 pillars in the proposal