Hours before tech CEOs were expected at the White House for a signing ceremony, President Donald Trump killed a proposed executive order on artificial intelligence Thursday—claiming he didn't want anything slowing America's lead over China in the AI race. "We're leading China, we're leading everybody, and I don't want to do anything that's going to get in the way of that lead," Trump told reporters at the White House.

What The Order Would Have Done

The axed directive would have created a framework for vetting advanced AI systems for national security risks before their public release. According to people familiar with the deliberations, it was designed as a voluntary collaboration with U.S.-based companies including Anthropic, OpenAI, and Google—tech giants that have all built frontier models now capable of finding vulnerabilities in existing software. The timing couldn't be more awkward: Treasury Secretary Scott Bessent and outgoing Federal Reserve Chair Jerome Powell had just convened an urgent April meeting with Wall Street CEOs specifically to warn them about cybersecurity risks from Anthropic's Claude Mythos model.

Internal Fractures On Display

Brown University computer science professor Serena Booth, a former AI policy fellow in a Democratic-led Senate committee, said the whiplash reflects deeper divisions inside the administration. "We do see this kind of public fighting," she noted. "'We will release an executive order. No, we won't. We're going to sign it this afternoon. Oh, the signing is canceled.' I think this whiplash is because we're seeing these fractures." The competing pressures include balancing reasonable pre-release testing for powerful models against concerns that government scrutiny could burden AI developers and slow innovation.

The Anthropic Problem

Complicating any government-industry cooperation on cybersecurity: the Trump administration has already banned Anthropic's Claude chatbot from federal agencies. In February, the president ordered all U.S. agencies to stop using the tool after an unusually public clash between the Pentagon and CEO Dario Amodei. That's a significant problem given that Treasury was just warning banks about Claude Mythos' capabilities—and now there's no clear path for those same institutions to access AI tools that might help defend against such models.

Commerce Department's Disappearing Act

The Commerce Department announced earlier this month it had signed agreements with Google, Microsoft, and Elon Musk's xAI to evaluate their most powerful AI models before public release—building on similar deals the Biden administration struck with Anthropic and OpenAI. Except that announcement later vanished from the Commerce website entirely, raising questions about what other commitments are quietly being walked back.

Key Takeaways

  • Trump killed a voluntary AI safety framework hours before signing ceremony to protect America's competitive edge
  • Treasury Secretary Bessent had warned banks about Claude Mythos' ability to find software vulnerabilities in April
  • The administration has banned Anthropic's tools from federal use while simultaneously warning industry about their power
  • Commerce agreements with Google, Microsoft, and xAI for pre-release model evaluation also seem to have disappeared

The Bottom Line

Let's be real here: powerful AI systems can already poke holes in financial infrastructure, and we're essentially hoping the companies building those systems volunteer to tell us about it. Trump's instinct to avoid heavy-handed regulation has merit—but canceling even a voluntary framework while China's own AI sector barrels forward leaves America's most critical systems exposed. The administration needs to pick a lane.