Trump Delays AI Security Executive Order, Saying Language Could Have Been a Blocker

President Trump delayed an executive order requiring pre-release government security reviews of AI models, saying its language could have been a blocker to US AI leadership.

Thursday May 21, 2026 Source: techcrunch.com
TL;DR — Quick Answer

President Trump delayed signing an AI security executive order on May 21, 2026 that would have let the government evaluate AI models before release, telling the White House press pool he did not like certain aspects of the language because it could have been a blocker to keeping the US ahead of China. The order would have tasked the Office of the National Cyber Director and other agencies with developing a pre-release security review process, partly in response to the release of Anthropic's Mythos and OpenAI's GPT-5.5 Cyber models. A key sticking point was a proposed requirement for companies to share advanced models with the government 14 to 90 days ahead of launch, and several outlets reported the real delay was that not enough tech CEOs could attend on short notice.

Key Takeaways

Trump Delays AI Security Executive Order, Saying Language Could Have Been a Blocker — AI news article illustration

President Donald Trump delayed signing an executive order on May 21, 2026 that would have allowed the government to evaluate AI models for security before they are released. “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,” he told the White House press pool.

A Delayed Signing

Trump said he was not happy with the order’s language, noting that a phrase in it could have been a blocker. Unofficial reports pointed to a different reason: not enough top tech CEOs could make it to Washington, D.C. on short notice — and, as one reporter put it, an executive order signing without a photo op is not a signing at all.

What the Order Would Have Done

The anticipated executive order would have tasked the Office of the National Cyber Director and other agencies with developing a process to evaluate AI models for security before their release. The push was partly a response to the release of Anthropic’s Mythos and OpenAI’s GPT-5.5 Cyber — both models that can quickly find and exploit security vulnerabilities.

The 14-to-90 Day Question

A key sticking point in the order’s language, according to CNN, was a proposed requirement for AI companies to share advanced models with the government between 14 and 90 days ahead of launch. That window would give federal agencies time to assess security risks before a model reached the public — but the industry warned it would slow release cadences and cede ground in the global AI race.

The Photo Op Problem

Multiple outlets reported that the real driver of the delay was logistical: too few invited tech leaders confirmed they could attend the signing ceremony on short notice. The administration, which treats such ceremonies as public displays of industry support, preferred to wait until it could stage the moment with the CEOs in the room.

What This Means

The delay leaves pre-release security reviews for frontier models in limbo — not cancelled, but not signed. For safety advocates it is a setback; for the AI industry it is a reprieve that keeps the US releasing models on its own schedule. What the administration does with the language, and who shows up for the eventual signing, will determine how binding the review process becomes.

Frequently Asked Questions

Why did Trump delay the AI security executive order?

President Trump said he was not happy with the language of the order, telling reporters that a phrase in it could have been a blocker and that he wanted to keep the US leading China in AI.

What would the delayed executive order have done?

It would have tasked the Office of the National Cyber Director and other agencies with developing a process to evaluate AI models for security before their release to the public.

What did the order require of AI companies?

A proposed requirement would have had AI companies share advanced models with the government between 14 and 90 days ahead of launch so its security risks could be assessed.

What models prompted the executive order?

The order was partly a response to the release of Anthropic's Mythos and OpenAI's GPT-5.5 Cyber, both of which can quickly find and exploit security vulnerabilities.

This article is based on the official announcement from techcrunch.com . Read the original for full technical details.

Related Articles

Back to all news