Remember? Just two weeks ago, Trump pulled his own AI executive order at the last minute, worried it might become a ‘blocker’ for the industry. Now he’s signed it — and the compromise is classic Trump: voluntary, but with clear expectations.
What’s In It
The Executive Order, titled ‘Promoting Advanced Artificial Intelligence Innovation and Security,’ has three pillars:
Voluntary model review: AI companies are asked to give the government early access to their most powerful frontier models — up to 30 days before public release. No mandate, no licensing requirement. But ‘voluntary’ in Washington carries its own weight.
AI Cybersecurity Clearinghouse: A new body will identify and patch vulnerabilities in AI systems at scale, working with industry and critical infrastructure operators — including rural hospitals and local utilities.
Criminal enforcement: The Attorney General is directed to prioritize prosecution of crimes involving AI for hacking, identity theft, or fraud.
What Stands Out
The order explicitly states it doesn’t create any mandatory licensing or pre-clearance requirement for AI models. That’s a direct shot at the Biden administration’s approach, which Trump calls ‘top-down regulation.’
At the same time, it establishes a classified benchmarking process for ‘advanced cyber capabilities,’ led by the NSA, CISA, and NIST. It reads as voluntary — but carries the unspoken subtext of ‘we’ll remember who doesn’t participate.‘
What It Means
For Anthropic, OpenAI, and Google, this means the government wants early access to new models. Voluntarily. But in an industry that depends on government contracts and goodwill, ‘voluntary’ has its own gravity.
The approach is more pragmatic than Biden’s 2023 executive order, which Trump revoked in January 2025. Less bureaucracy, more partnership — at least on paper.
Sources: White House Fact Sheet, CNBC, NPR