Anthropic CEO Dario Amodei published an open letter this week calling to “pace the frontier” and slow AI development. OpenAI’s Sam Altman agreed. So did Elon Musk. Even Google DeepMind’s Demis Hassabis offered tentative support.

President Trump and House Speaker Mike Johnson aren’t buying it.

The administration’s response marks the first major policy split between AI executives and federal leadership since Trump took office. Both Trump and Johnson dismissed the industry’s safety concerns as overreaction, according to statements reported by The Verge. Their reasoning centers on competition with China.

The concern isn’t new. It’s been the default counterargument to AI safety proposals since GPT-4 launched: slow down and China wins. What’s different is hearing it from the White House while the people actually building the systems are asking for guardrails.

The pause proposal

Amodei’s letter doesn’t call for stopping AI research entirely. “Pacing the frontier” is deliberate language. It suggests keeping development at a rate where safety research can keep up with capabilities research. The idea is that racing to build more powerful systems without understanding how to control them creates unnecessary risk.

This isn’t Amodei’s first time raising safety concerns. Anthropic was founded in 2021 by former OpenAI researchers who left over disagreements about the company’s approach to safety. The company has consistently positioned itself as the more cautious alternative in the AI race.

What is notable is who else signed on. Altman spent years arguing that slowing down wasn’t realistic. Musk has been all over the map on AI safety, but his companies continue building AI systems at full speed. Hassabis has been more measured, but Google is still pushing hard on AI capabilities.

Getting all three to publicly support a pause, even a temporary one, suggests something shifted. The question is what.

The China argument

Trump and Johnson’s response relies on the assumption that China won’t slow down if the US does. That’s probably true. China’s AI development has been explicit about national strategic goals. The government funds research, directs deployment, and doesn’t appear particularly concerned with the safety questions worrying Western researchers.

But the China argument has always had a problem: it assumes the only thing that matters is who builds the most powerful AI first. It doesn’t address what happens if either country builds something they can’t control.

There’s no legal framework yet for what an AI pause would look like. Export controls limit chip sales to China, but those focus on hardware, not model development. The White House could potentially restrict research funding or impose safety requirements on federally funded AI projects. Neither would stop private companies from continuing their work.

Congress hasn’t shown much interest in mandatory AI safety standards. Johnson’s comments suggest that’s unlikely to change while Republicans control the House. Democrats have been more receptive to regulation, but former President Obama’s recent comments suggest even Democratic leadership is focused more on economic impact and job displacement than existential risk.

What happens next

Nothing, probably. Tech executives calling for regulation while continuing to build the thing they want regulated is a familiar pattern. Facebook did it with content moderation. Google did it with privacy. The playbook is to ask for rules, keep building while waiting for those rules, and then lobby to weaken whatever eventually gets proposed.

The difference with AI is the executives seem genuinely uncertain about what they’re building. That uncertainty hasn’t slowed development, but it’s created this strange dynamic where the people racing to build AGI are also the ones warning it might be dangerous.

Whether that concern is genuine or strategic doesn’t really matter for policy purposes. Without federal action, there’s no mechanism to actually pause AI development. The companies can slow themselves down if they want to. Based on their hiring, their chip orders, and their product announcements, they don’t want to.

Trump and Johnson’s response makes clear the federal government won’t force them to. The White House position is that the race matters more than the risks. That might be the right call. But it’s a bet, not a certainty, and we won’t know if it pays off until it’s too late to change course.