AI Leaders Call for a Slowdown While Insisting the Race Must Continue

AI Leaders Call for a Slowdown While Insisting the Race Must Continue
The AI industry’s fiercest competitors have found rare common ground: development is moving too quickly. But their agreement ends there, with leaders divided over whether voluntary safeguards can manage the danger without sacrificing technological and strategic advantage.

Anthropic CEO Dario Amodei’s proposal to “pace the frontier” would give independent evaluators employee-level access inside AI companies, encourage shared safety standards among democratic nations and seek basic safeguards even with authoritarian governments. He argues that progress should continue, but with enough time for testing and alignment to catch up.

OpenAI CEO Sam Altman endorsed the plan, saying independent evaluators were “a great idea,” while Elon Musk offered a blunt vote of confidence: “Dario is right.” Their support marks a striking convergence among rivals whose businesses depend on pushing model capabilities forward. OpenAI’s own chief scientist has warned that no company has solved alignment and monitoring well enough to keep scaling “at maximum speed.”

Yet the proposed slowdown is deliberately limited. Amodei says “pacing” does not mean halting training or technical progress, and he has warned that moving too slowly could allow autocratic governments to take control of the technology. That strategic concern aligns with a harder-edged U.S. position from Rep. Anna Paulina Luna, who called advanced AI a “massive — uncontrolled currently — super weapon” and said America must still win the race.

Critics from the safety camp see voluntary commitments as insufficient. Former OpenAI researcher Daniel Kokotajlo welcomed Amodei’s proposal as “a step in the right direction,” but said it may not go far enough. Former Anthropic researcher Jacob Coxon has likewise called for transparent third-party auditing, warning that continued competition could make genuinely reliable safety assurances impossible.

The result is a consensus with a built-in contradiction: the companies accept that unchecked acceleration is dangerous, while governments and executives remain unwilling to surrender the race that acceleration is driving.