The AI industry’s most powerful executives just asked the world to pump the brakes. Donald Trump and Mike Johnson aren’t interested.

On Friday, Anthropic CEO Dario Amodei published an open letter calling for the industry to “pace the frontier” and slow down AI development. Within hours, OpenAI’s Sam Altman and Elon Musk both voiced support on X. Even Alphabet’s Demis Hassabis offered tentative backing.

By Saturday, Trump and House Speaker Mike Johnson had their response: no. According to The Verge, both believe the AI executives are overreacting. Their concern isn’t safety. It’s China.

This isn’t the first time the industry has called for a pause. In March 2023, the Future of Life Institute published an open letter asking all AI labs to halt training of systems more powerful than GPT-4 for at least six months. That letter got thousands of signatures, generated weeks of debate, and changed precisely nothing. Labs kept training bigger models.

But this time feels different. The support isn’t coming from academics and researchers on the sidelines. It’s coming from the people actually building the systems. Altman runs the company that kicked off the current AI race with ChatGPT. Musk owns xAI, which just raised billions to compete with OpenAI. Amodei left OpenAI to start Anthropic specifically because he thought the company wasn’t taking safety seriously enough.

When these three agree on anything, it’s worth paying attention.

The timing matters too. We’re now three and a half years into the current wave of AI development. The low-hanging fruit has been picked. Models are getting more expensive to train, and the gains are getting harder to measure. Multiple reports suggest that GPT-5 and other next-generation models haven’t delivered the leap forward their developers expected.

At the same time, the infrastructure demands keep growing. A WIRED report this week highlighted how the shift from chatbots to autonomous AI agents is driving massive data center expansion. These agents don’t just answer questions. They take actions, run continuously, and consume far more computing resources than a simple query to ChatGPT.

The economics are starting to look questionable. The safety concerns are mounting. And now the people running the labs are saying maybe we should slow down and figure out what we’re doing.

Trump and Johnson’s response cuts straight to the political reality that’s shaped AI policy since the beginning: fear of losing to China. That fear has been the industry’s most effective argument against regulation. Every time someone proposes new safety requirements or restrictions on AI development, the response is the same. If we slow down, China wins.

It’s not entirely wrong. China is investing heavily in AI. Chinese labs are producing competitive models. And the Chinese government has shown it’s willing to use AI for surveillance and control in ways that would be politically impossible in the US.

But the argument has also become a convenient way to shut down any conversation about tradeoffs. Should we require safety testing before deploying powerful AI systems? Can’t do that, China will win. Should we limit access to the most advanced chips? Too risky, China will catch up. Should we maybe spend six months figuring out if we’re building something dangerous? Absolutely not, China doesn’t pause.

The fact that Trump and Johnson are making this argument now, when the industry’s own leaders are asking for a slowdown, shows how deeply it’s embedded in the political conversation around AI.

There’s also a credibility problem. The AI industry has cried wolf about existential risk so many times that it’s hard to know when to take the warnings seriously. These same executives have spent years saying AI might kill everyone while simultaneously racing to build more powerful systems and lobbying against meaningful regulation.

Obama weighed in this week too, urging Democrats to make AI a “central agenda” and develop a “clear plan” to address economic and safety concerns. That’s the kind of vague statement that sounds reasonable but commits to nothing. What does a “clear plan” look like? What specific policies would it include? Obama didn’t say.

The pause debate is useful for one thing: it shows that even the people building these systems don’t have a clear picture of where this is heading or how to control it. That’s not an argument for stopping. But it might be an argument for having some actual answers before we scale up further.

For now, the industry will keep building. Trump and Johnson’s opposition probably kills any political momentum for a pause, at least in the US. And if the US doesn’t pause, no one else will either.

The question isn’t whether the AI executives are right about the risks. It’s whether “China might beat us” is a good enough reason to ignore them.