President Trump on Tuesday signed an executive order establishing a voluntary framework for government review of border artificial intelligence models before their public release, ending weeks of internal conflict in the White House over how aggressively to regulate the technology. The order, titled “Promoting innovation and security of advanced artificial intelligence,” was signed privately without the usual live broadcast or public ceremony, in contrast to the fanfare that normally accompanies AI presidential announcements.
The final version is substantially narrower than the draft that Trump rejected on May 21, when he canceled a planned signing ceremony over fears that the order “could blunt the United States’ advantage in artificial intelligence technology.” The original draft proposed a mandatory 90-day pre-launch review period. and would have given the government formal evaluation authority over frontier models. The signed version asks companies to voluntarily submit models 30 days before launch and participate in a collaborative framework instead of undergoing mandatory testing.
what makes the order
The executive order establishes three main mechanisms. First, a voluntary pre-launch review framework in which AI developers can engage the government to determine whether models in development qualify as “covered border models“, provide access up to 30 days before the planned launch and collaborate on the selection of “trusted partners” for early access. The framework is explicitly voluntary, meaning companies can refuse to participate without penalty.
Second, the order creates an AI cybersecurity clearinghouse within 30 days, coordinated by the Secretary of the Treasury, the National Cyber Director, the NSA, and CISA. The clearinghouse will search for software vulnerabilities, validate discoveries, and coordinate remediation and patch distribution. a direct response to the Myth crisis That demonstrated how vulnerabilities discovered by AI can overcome existing disclosure and patching processes.
Third, federal agencies should develop benchmarks to evaluate the cybersecurity capabilities of AI models and strengthen the government’s own security defenses against AI-enabled threats. The order also addresses AI safety research.although the specific provisions are less prescriptive than those contained in the original project.
that was cut
The differences between the discarded draft and the signed order reflect the victory of the pro-industry faction within the White House. The mandatory 90-day review was reduced to a voluntary 30-day window. Formal government evaluation authority was replaced by a collaborative framework. Reporting requirements for companies developing powerful models, which would have echoed provisions of Biden’s repealed AI executive order, were relaxed to avoid what industry allies characterized as regulatory overreach.
Silicon Valley’s objections to the original draft They were decisive. AI companies argued that mandatory pre-launch testing would slow American innovation, create a competitive disadvantage relative to Chinese companies that do not face equivalent requirements, and set a precedent for government control of technology deployment. The signed order addresses those concerns by making participation voluntary and framing the government’s role as collaborative rather than regulatory.
The hole it leaves
The voluntary framework means that the effectiveness of the order depends entirely on whether AI companies decide to participate. Companies already engaged in pre-launch testing with CAISI, including Google, Microsoft, and xAI, may continue or expand that cooperation. Companies that find government review commercially disadvantageous or rush to ship products can simply opt out.
EU AI Law to come into force in Augustoffers a stark contrast: mandatory requirements, legal authority, and penalties for noncompliance. Trump’s order sets standards and creates institutional infrastructure (the cybersecurity clearinghouse, the benchmark development process), but it is based on goodwill rather than obligation.
For the White House, the silent signature may be the point. The order gives the administration a policy document it can refer to when asked about AI oversight, creates structures that could be strengthened later, and avoids a public confrontation with an AI industry whose leaders are among the administration’s most visible supporters. Whether a voluntary framework is appropriate for a technology that can discover 10,000 zero-day vulnerabilities in a month is the question the order deliberately leaves unanswered.
