On May 21, Donald Trump was supposed to sign an executive order on AI regulation in the Oval Office. The pens were ready, the cameras set up. Then the ceremony was called off — just hours before.
What Was Planned?
The executive order would have created a voluntary framework for AI companies to share frontier models with the US government up to 90 days before public release. The goal: catching cybersecurity risks earlier.
Sounds like a reasonable compromise. But that’s exactly where it fell apart.
Why the Order Failed
According to reports from CNN and the Washington Post, two irreconcilable factions had formed inside the White House. On one side: the pro-innovation camp, which wanted the framework to remain genuinely voluntary. On the other: national security advisors who demanded NSA-backed classified testing with real enforcement teeth.
Trump himself put it simply: “I didn’t like certain aspects of it. I think it gets in the way — you know, we’re leading China, we’re leading everybody, and I didn’t want to do anything to get in the way of that lead.”
What This Means
The cancellation is telling. It shows how hard it is to regulate AI when your own government can’t agree on what it actually wants.
For Anthropic, OpenAI, and Google, nothing changes in the short term — there’s no new obligation. But the uncertainty remains. Nobody knows if or when a revised order will come. And the conversation about frontier model testing isn’t going away.
The ironic punchline: just weeks ago, the White House announced an FDA-style AI review process. Now it can’t even agree on the much softer version.
Sources: CNBC, Axios, Washington Post