What's behind Washington's AI safety pivot

The Trump administration appears poised to reshape the U.S. approach to AI security ahead of President Trump's trip to China next week.

Why it matters: What happens next could be the turning point for how the Trump White House handles the proliferation of the most advanced AI models in the world.

• And there are new reports of possible coordination between the two countries that are fiercely competing on AI development — a signal that neither side wants a dangerous arms race.

Driving the news: A fire-alarm moment is happening:

• The pro-AI growth administration is realizing it may need more guardrails than originally thought, and may not want to go it alone.
• There are new signs that the administration may consider executive action to rein in the most powerful AI models.
• At the same time, the U.S. and China are weighing official discussions about AI, and it could be added to next week's Beijing summit between Trump and Chinese leader Xi Jinping, the Wall Street Journal reported this week.

What they're saying: National Economic Council director Kevin Hassett suggested this week that the administration is considering an executive order, hinting at an oversight process for new AI models that would be similar to Food and Drug Administration approval of new drugs.

• "We're studying, possibly an executive order to give a clear roadmap to everybody about how this is going to go and how future AIs that also potentially create vulnerabilities should go through a process so that they're released to the wild after they've been proven safe, just like an FDA drug," Hassett told Fox Business on Wednesday.

White House chief of staff Susie Wiles also weighed in with a more general statement on X Wednesday night:

• "When it comes to AI and cyber security, President Trump and his administration are not in the business of picking winners and losers. This administration has one goal; ensure the best and safest tech is deployed rapidly to defeat any and all threats," Wiles wrote.
• "We appreciate the effort being made by the frontier labs to ensure that goal is met."

Zoom in: Wiles mentioned "safety" three times in her post. Compare that to Vice President JD Vance's comments on AI in February 2025 at the AI Action Summit in Paris:

• "We need international regulatory regimes that foster the creation of AI technology, rather than strangles it," he said. "The AI future is not going to be won by hand-wringing about safety."

The latest: The government appears to be mulling a number of executive actions to possibly announce before Trump goes to China, sources tell Axios, cautioning that all talks are in flux and nothing is final.

• As Axios has been reporting, the possible measures include an executive action focused on AI and cybersecurity; one related to deployment and testing of new AI models; and another that could be some form of licensing or approval around limitations a model provider could place on government use of AI.
• This week, White House meetings have included both tech and financial services companies, one source familiar with the discussions told Axios, with Treasury Secretary Scott Bessent wanting banks to be looped into whatever happens.
• Google, xAI and Microsoft also signed pre-deployment testing deals this week with the Center for AI Standards and Innovation, part of the Department of Commerce, and announced continued deals with Anthropic and OpenAI.

The other side: "The White House continues to balance advancing innovation and ensuring security in our AI policymaking. The Chief of Staff's X post reiterated this longtime commitment," a White House official said.

Reality check: A rhetorical shift is just that until the administration announces concrete steps beyond this week's hints.

• Recent statements "reflect less of a major policy shift than an internal debate about how to reconcile its commitment to limit AI regulation with the realities of AI progress," Vivek Chilukuri, senior fellow at Center for a New American Security, told Axios.
• "The true test will come when a frontier lab ... forges ahead with a model release even if the government finds credible security risks."
• One former Commerce staffer told Axios: "It feels like they got spooked with Mythos and realized that, 'Oh shit, we might actually need to do something.' But what are they actually doing that's new?"