Trump Pulls AI Safety Order Hours Before Signing After Big Tech Pushback

Welcome back to SavvyMonk, your one-stop for AI and tech news that actually matters.

Today we're looking at how the most significant AI regulation attempt of the Trump administration fell apart in a matter of hours, and what it tells us about who actually controls the rules of the AI race.

Let's get into it.

A year ago, most marketers weren't thinking about AI search. Now it's one of the fastest moving channels in the industry and nobody has a playbook yet.

So we built one. We surveyed hundreds of marketers to find out how they're approaching answer engine optimization, where they're investing, what's actually working, and what isn't.

The result is the 2026 State of AEO Report. Real data. Real strategies. A clear picture of where AI search is headed and how to get ahead of it.

Download the free report

On May 21, 2026, President Trump was set to sign a highly anticipated executive order that would have created a first-of-its-kind system for the federal government to vet powerful AI models before they reach the public.

The White House had already sent invitations to executives from leading tech companies. The signing ceremony was scheduled for the afternoon.

But Trump pulled the plug before the cameras even turned on.

"I didn't like certain aspects of it," he told reporters in the Oval Office. "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 leading."

The unofficial reason was far less philosophical. Not enough tech CEOs could make it to Washington on short notice, and apparently you don't sign a landmark AI order without a proper photo op.

What the Order Would Have Done

The proposed executive order would have tasked the Office of the National Cyber Director and other federal agencies with building a process to evaluate AI models for security vulnerabilities before their public release. This wasn't a theoretical exercise, and it came as a direct response to the release of two specific models that rattled the national security establishment.

Anthropic's Mythos and OpenAI's GPT-5.5 Cyber have both demonstrated the ability to quickly find and exploit security flaws in computer code. Those capabilities made the case for some form of pre-release review almost impossible to ignore, even for an administration that has championed a hands-off approach to AI.

One of the biggest sticking points was a proposed requirement for AI companies to share their advanced models with the government between 14 and 90 days before launch, a provision Trump warned could block the pace of U.S. AI development.

Who Stopped It

Mark Zuckerberg, Elon Musk, and former White House AI and crypto czar David Sacks all spoke with Trump between Wednesday night and Thursday morning, and by the time the sun came up, the executive order was effectively dead.

The lobbying had been building for days after the Office of the National Cyber Director briefed AI companies on the proposed model-review framework, but those overnight conversations sealed it.

The internal politics were surprisingly messy. Some Trump allies, including Steve Bannon and other MAGA-aligned figures, had actually pushed for tougher AI oversight on national security grounds, while major tech companies argued that advance government review would slow U.S. development and give China an edge.

Silicon Valley won that argument before the White House staff even got to their desks.

The Bigger Picture

This was supposed to be a pivot moment. When Trump entered office, he revoked the Biden administration's landmark AI executive order, which had required tech companies to notify the government when building advanced models and share their safety evaluations with the Commerce Department. The new order was meant to replace that framework with something more focused on cybersecurity rather than broad safety testing.

The fact that even this narrower approach got shelved tells you everything about the current balance of power between Washington and the AI industry. The administration wanted to move fast, reportedly giving agencies about two months to build the review framework. But when the companies pushed back, the timeline collapsed.

For now, AI companies will continue launching frontier models without any formal government security review. The Mythos and GPT-5.5 Cyber capabilities that triggered this whole effort are still out there, and the policy vacuum they exposed remains wide open.

The Bottom Line

The people building the most powerful AI systems on the planet convinced the President of the United States to shelve the only mechanism that would have given the government a look at those systems before they go live.

Whether you think that's a win for innovation or a failure of governance probably depends on which side of the table you're sitting on. But the precedent is set, and the industry now knows exactly how much leverage it has.

Category: Policy Analysis

"Summarize the key differences between the Biden-era AI executive order (October 2023) and the proposed Trump AI security executive order (May 2026). Focus on what each required from AI companies, which agencies were responsible, and the enforcement mechanisms. Present it as a comparison table with columns for each order."

There's something worth sitting with here. An administration that has positioned itself as tougher on China than any in recent memory backed down from an AI security measure because the companies told them it would slow things down.

The same companies, by the way, that are spending hundreds of billions on AI infrastructure this year. The question isn't whether AI needs oversight.

It's whether anyone in Washington has the leverage to actually do it.

Hit reply, I read every response.

See you in the next one.

— Vivek

P.S. If you know someone who follows AI policy and wants the full picture without the spin, forward this their way. They can subscribe at https://savvymonk.beehiiv.com/