On 21 May 2026, a few hours before a planned signing ceremony, the White House postponed a significant AI executive order.
The suggested order aimed to establish an optional system for AI companies to share advanced models with the federal government up to 90 days before public release – a modest early-warning system, explicitly non-mandatory and carrying no licensing or approval requirements. It was about as gentle a piece of AI oversight as you could design.
It still didn’t make it to the President’s desk – last-minute calls from Elon Musk, Mark Zuckerberg and David Sacks – Trump’s own AI and crypto adviser – pushed back hard enough that Trump declined to sign. His stated reason: “I believe it interferes with our lead over China, our lead over everyone, and I don’t want to do anything that might hinder that lead.” Both Musk and Meta have since denied being the deciding factor. OpenAI, for what it’s worth, supported the order.
Whatever the internal politics, you get the same result: the US now has no formal plan or timeline to manage risks from advanced AI systems. And for companies building AI products, that matters more than the drama surrounding how it happened.
What The Regulatory Vacuum Actually Means
The instinct might be to read a collapsed AI regulation as good news for founders – less oversight, more room to move – that reading is wrong. The absence of a federal framework doesn’t create clarity, it creates fragmentation.
Fifty US states now have their own AI laws, which took effect on 1 January 2026. Without federal preemption, any company building and distributing AI products across the US faces a patchwork of overlapping, sometimes conflicting rules depending on where it operates or incorporates. Rather than relying on a federal standard to streamline compliance, companies are now forced to handle a fragmented, state-by-state patchwork on their own. That’s a real cost for early-stage startups that don’t have legal teams built for it.
There’s also the international dimension – the EU AI Act becomes fully applicable on 2 August 2026 – less than ten weeks away. For businesses operating in or selling to the EU, those obligations are live regardless of what Washington does or doesn’t do. The UK is running a sector-specific, agile approach with no central AI regulator. And the US is now, as one analyst put it, “further behind Europe and Asia on AI governance than at any point in the past three years.” The maze of international regulations is becoming more intricate by the day.
More from Artificial Intelligence
- Nvidia Just Posted $81 Billion In Revenue – And The Gulf Is One Of Its Biggest Customers. Is That A Risk?
- Thrive Unveils Its Most Intelligent Release Yet, Putting A Team Of AI Agents At The Heart Of Every Employee Experience
- VEED Launches Subtitle API To Power The Next Generation Of AI Content Platforms
- Experts Share: Is AI The Next Big Excuse Behind Another Wave Of Mass Tech Layoffs?
- AI Shopping Tools Are Taking Over, But Consumers Still Trust Known Brands More
- Introducing Halupedia: The Wikipedia Clone Made Up Of AI Hallucinations
- Europe Is Building The AI Infrastructure And The Harder Part Is Making It Work
- Google Just Had Its Most Ambitious I/O Ever – Here Is The Part That Affects You
The Accelerationist Argument Won – For Now
The argument Musk, Zuckerberg and Sacks made to Trump was essentially the accelerationist case: any friction between AI companies and government slows the US down in its competition with China, and that risk outweighs the risk of moving without protections. National Economic Council officials were reportedly sympathetic to the same position.
That argument has won the immediate battle – whether it holds depends on what happens next. The Anthropic–Pentagon dispute over autonomous weapons showed that the relationship between AI companies and government isn’t straightforwardly cooperative even when regulation is off the table. OpenAI is now pursuing state-level regulations with White House approval, which means the regulatory battlefield is shifting rather than disappearing. Founders will face uneven rules depending on where they incorporate or operate – which is exactly the kind of uncertainty a coherent federal framework was supposed to prevent.
Adapting to the Global Landscape
Anyone operating outside the US needs to pay close attention to this week’s developments. American AI regulation is further away than ever, which means US-based competitors are operating without the compliance costs that EU AI Act obligations will impose on European companies from August. That’s an uneven playing field you must strategize for, rather than assuming it will level out on its own.
The more durable lesson is about where regulatory certainty actually exists right now. The EU has a framework, whatever its imperfections. The Gulf – through VARA, ADGM and the UAE’s Payment Token Services rules – has built clearer AI and fintech governance than most Western markets. The UK has a light-touch approach that, for now, gives AI companies room to operate without the compliance weight of the EU.
The US, by contrast, has a charismatic vacuum: lots of activity, no settled framework, and an unpredictable relationship between the government and the companies it’s simultaneously depending on and failing to regulate. Three phone calls on a Thursday afternoon made that clearer than any policy paper could.
