Donald Trump's AI Executive Order: What Changed After Delayed Announcement

President Donald Trump privately signed an executive order on Tuesday that he said was seeking to regulate artificial intelligence (AI) without stifling growth, days after an original order was pulled at the last minute.

Trump wrote that while AI was a useful new tool that will make the United States stronger, it also posed a national security threat and needed government oversight, although this appeared to have been scaled back after tech companies were said to have convinced the president to hold off on his previous plan.

The order issued Tuesday calls for technology companies to submit new AI models for government review 30 days before release, a significant cutback from the reported 90-day window originally planned.

AI Executive Order: What it Says

The order tells federal agencies to move quickly to harden government and critical infrastructure systems against AI-related cyber risks, create a voluntary process for the government to review the most advanced models before wider release, and step up criminal enforcement against people using AI to hack or commit related crimes.

“My Administration has unleashed tremendous technological growth and economic investment in AI by slashing the bureaucratic constraints that the prior administration placed on America's AI developers and researchers, and by instead encouraging AI innovation and accelerating responsible AI adoption across government and industry,” the order read.

The practical effect is that the administration is trying to do two things at once: avoid broad new AI regulation while still building a federal mechanism for dealing with the cybersecurity risks of the most advanced models.

For consumers, the order does not suddenly create a comprehensive federal AI law but tries to prepare critical infrastructure for any potential security issues the technology creates.

The order protects banks, hospitals and other critical infrastructure indirectly through cybersecurity measures, not by regulating how they use AI day‑to‑day. The key idea appears to be to make the systems they rely on harder to hack or disrupt as AI-powered attacks become more sophisticated.

For agencies, critical infrastructure operators and developers of the most capable models, the order sets short deadlines, creates a clearinghouse for new models and opens the door to structured pre-release collaboration with the government.

What Did Trump Change?

The order appeared less aggressive than what was being publicly discussed a few weeks ago.

In May, Politico reported that the administration was considering an executive order that could create a vetting regime for frontier AI models, and that the process could require companies to get a government "green light" before releasing advanced models.

The same report said National Economic Council Director Kevin Hassett confirmed that the White House was studying a possible order that would allow the administration to block release of AI models deemed unsafe.

Some companies, including OpenAI, were said to be in favor of the idea, while others, including Elon Musk and Meta’s Mark Zuckerberg, were said to have reservations, warning that the planned vetting would have damaged the economy.

The June 2 order stops short of this. While it creates a benchmark for "covered frontier models" and a pre-release government access framework, it says the framework is voluntary and then goes out of its way to say it does not authorize mandatory licensing, preclearance or permitting.

Compared with the tougher ideas floated in May, the signed order looks like a compromise version: the government still gets a structure for evaluating highly capable models but not a formal approval gate before release.

What Happens Next

Over the next 30 days, federal agencies will roll out cybersecurity guidance, stand up a vulnerability clearinghouse, and expand protections for critical infrastructure like hospitals and banks. Within 60 days, officials must define which AI systems qualify as "frontier models" and begin building a voluntary process for companies to share those systems with the government before release.

After that, the impact will depend on how aggressively agencies implement the rules and whether AI companies choose to participate in what remains a voluntary framework rather than a mandatory regulatory regime.

2026 NEWSWEEK DIGITAL LLC.

This story was originally published June 2, 2026 at 10:01 AM.