The White House called off a planned signing ceremony for President Trump’s AI executive order on Thursday afternoon, just hours before it was scheduled to happen. Trump told reporters in the Oval Office that he pulled back because he “didn’t like certain aspects of it” and feared the order “could have been a blocker” to US AI innovation.

“We’re leading China, we’re leading everybody,” Trump said, “and I don’t want to do anything that gets in the way of that lead.”

Major tech, AI, and cybersecurity company CEOs had been invited to attend the ceremony. They left without a signed order.

What the Order Would Have Done

The executive order centred on AI security and pre-release model review. Its core provision was a voluntary framework giving the US government up to 90 days to evaluate new AI models before public release. Under the proposal, major AI developers including OpenAI, Anthropic, and Google would have shared advanced models with federal agencies during a pre-release window.

The cybersecurity section outlined a voluntary “clearinghouse” run by the Treasury Department alongside other federal agencies and AI companies. The goal was to identify and fix security vulnerabilities in unreleased models before they reached the public. The order also called for expanded hiring at the US Tech Force, the government body of engineers working to modernize federal computer systems.

Even in its final form, participation was entirely voluntary. Companies could join the review process or skip it.

A Pattern of Delays

This postponement follows a longer pattern. The signing ceremony had already been rescheduled multiple times before Thursday’s last-minute cancellation. Earlier this month, White House economic adviser Kevin Hassett publicly described the administration as studying an FDA-style vetting process for AI models, language that suggested a more mandatory framework. The final draft had already dialled back considerably from that description, and now even that version has been shelved.

The order’s iterative drafting process reflects a genuine tension at the heart of US AI policy: how do you protect national security without slowing down the industry you are trying to protect?

What the Business Community Expected

There had been real optimism around this order, not because businesses wanted more rules, but because voluntary guidance from the federal government gives large enterprises something to point to. Without it, every company building or deploying frontier AI models is effectively setting its own bar.

The clearinghouse concept was particularly interesting for enterprise buyers. A government-backed process for identifying security vulnerabilities in AI models before they go live would have helped corporate risk teams justify AI investment. It offered a credible third-party security signal at a time when enterprise security teams are still working out how to audit AI outputs and agent behaviour.

What This Means for Business

For companies currently deploying or planning AI systems, today’s postponement confirms that federal clarity on AI governance is not arriving soon.

The practical reality for businesses is unchanged from yesterday: there are no federal rules mandating how you develop, deploy, or audit AI systems. That is a double-edged position. It means freedom to move quickly. It also means no external floor forcing competitors to meet any baseline standard.

A few things are worth keeping in mind as this plays out:

The gap is your opportunity. Companies building internal AI governance practices now will not be scrambling when rules eventually land. That means documenting what your AI systems do, tracking outputs, establishing human review checkpoints, and training your team to work alongside AI tools responsibly.

Voluntary today does not mean voluntary forever. The administration has now postponed this order multiple times but has not abandoned it. The EU AI Act compliance clock is also running for any business with European operations. Federal rules will come; the debate is about timing and form.

Security posture matters regardless of regulation. The clearinghouse concept addresses a real problem. Whether or not the government formalises it, responsible AI deployment requires businesses to think about what happens when an AI system is wrong, misused, or manipulated. That is an internal governance question, not a compliance checkbox.

For businesses that have been waiting on Washington to define the rules before making AI decisions, the message from today is straightforward: the rules are not coming on a fixed schedule. Build your own governance framework and get moving.

---

If this is the kind of problem agents can help with, the free Working With Claude field guide is the practical next step. Thirty-two pages, no fluff. Get the free guide.