Pacing the Frontier: AI Leaders Push for a Strategic Slowdown

Anthropic CEO Dario Amodei has proposed a three-step plan to "pace the frontier" of AI development, calling for a reduction in the speed of model capability improvements. Amodei cited the rapid acceleration of AI's ability to build next-generation systems and a cybersecurity incident involving OpenAI and Hugging Face as primary drivers for this cautious approach.

As a first step, Amodei announced that Anthropic is "unilaterally committing" to allowing third-party organizations, such as METR, to embed evaluators within the company. These evaluators would receive company badges, laptops, and access comparable to internal risk teams to verify safety commitments. OpenAI CEO Sam Altman agreed with the proposal and stated that OpenAI would follow suit.

Amodei's second step calls for AI companies in democratic nations to coordinate common safety standards and limits on unchecked progress. He noted that the U.S. government would need to provide a narrow antitrust waiver to enable these safety discussions. For the third step, Amodei suggested global coordination with authoritarian governments, including China, to prohibit dangerous uses of AI, such as biological weapons production.

To maintain a technological lead over China during this slowdown, Amodei suggested the U.S. government restrict the sale of powerful chips and semiconductor equipment and crack down on model distillation.

Critics have questioned the motives behind these proposals. Journalist Brian Merchant told TechCrunch that such measures would likely only benefit the incumbents, describing the effort as "what regulatory capture looks like in action."