Trump signs scaled-back AI executive order after months of internal debate

From 90 days to 30: how the order was scaled back

The final version cuts the government's pre-release review timeline for advanced AI models from 90 days to 30 days, and it backs away from any mandatory vetting regime. Instead, the order relies on voluntary cooperation from the companies leading AI development, including OpenAI, Anthropic, and Google. The text makes that voluntary nature explicit, stating that 'nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models.'

Beyond the testing framework, the order directs federal agencies to develop benchmarks to assess AI models' cyber capabilities and to create an AI cybersecurity clearinghouse that would review and share information on vulnerabilities. It also instructs agencies to shore up the government's own security defenses, a response to growing concerns that adversaries could weaponize the most capable systems.

Internal divisions over the scope of regulation

The order reflects a Trump administration that has been divided over how to approach AI regulation. Trump has sought to minimize federal oversight, including action at the state level, even as some officials argued that more robust safeguards were needed. The Biden White House had pushed for federal oversight of the technology, and the new order is being read as a compromise between those two impulses.

The signing comes against a backdrop of heightened concern in Washington and Silicon Valley about the security risks of frontier AI systems. In April, Anthropic announced that it was limiting the release of its new Mythos Preview model because of its ability to identify and exploit software security vulnerabilities, an announcement that set off alarm bells across the industry and the capital. Pro-regulation voices are hailing the executive order as proof that momentum is on their side, even though the document stops short of imposing binding requirements.

Anthropic's Mythos model and the security backdrop

The order itself frames artificial intelligence as both a strategic asset and a security challenge. It states that 'Advanced AI capabilities make our Nation stronger, but also introduce new national security considerations that require coordinated action across executive departments and agencies.' A second passage adds that 'As these capabilities evolve, my Administration will continue to work closely with industry to ensure that the best and most secure technology is deployed rapidly to confront any and all threats to our country.'

For now, any binding rules on AI testing and deployment would have to come from Congress rather than the executive branch. The order is positioned as a coordination mechanism between government and industry, setting up benchmarks, a vulnerability clearinghouse, and a 30-day pre-release testing window, but leaving enforcement to voluntary participation by the leading developers.

Industry response in the immediate aftermath of the signing was not detailed in the available reporting, but the inclusion of OpenAI, Anthropic, and Google as the de facto partners for the voluntary program suggests the White House intends to anchor its AI policy around the largest U.S. model builders. Whether that framework holds will depend on how those companies respond to requests to share their most capable systems, and on whether Congress moves to translate the voluntary regime into law.