Trump signs scaled-back AI order for voluntary 30-day security reviews

President Donald Trump has signed a scaled-back executive order on artificial intelligence that would let the federal government review advanced AI systems for national security risks before release, but only on a voluntary basis and for a shorter, 30-day period, according to the Associated Press and the Council on Foreign Relations. The order came after Trump delayed an earlier version over concerns that stricter oversight could weaken the United States’ competitiveness with China.

The order creates a framework for federal agencies to examine the most advanced AI models before they are publicly released, with the goal of identifying security and reliability risks. But unlike a more forceful regulatory approach, participation by developers is not mandatory, and the review window was cut to 30 days, a change that appears designed to limit the burden on companies racing to ship new models, the AP reported. The Council on Foreign Relations said the revised order preserves much of the original structure while reducing the review period from the 90 days that had reportedly worried the administration in May.

The timing reflects the growing political and strategic importance of frontier AI systems. According to the reporting, Trump postponed a White House signing event in May after objecting to an earlier draft, and industry pressure helped shape the final text. The initial proposal had drawn objections from AI companies that feared a more aggressive review process could slow innovation, especially in a global race with China.

The order also comes as concern has intensified around powerful new models such as Anthropic’s Claude Mythos, which the sources say helped sharpen attention inside the administration on AI cybersecurity risks. The Council on Foreign Relations noted that the policy is aimed in part at systems capable of finding and exploiting hidden software vulnerabilities, a concern that has pushed governments to consider tighter scrutiny of frontier models.

According to the White House language quoted by the Council on Foreign Relations, the administration says the policy is meant to promote AI innovation and security while protecting intellectual property and guarding against threats from adversaries. That framing underscores the balancing act behind the order: Washington wants to avoid slowing a fast-moving industry while also responding to the national security risks that advanced AI is increasingly seen to pose.

The move leaves important questions unresolved, including how many companies will choose to take part in the voluntary reviews and how aggressively federal agencies will use the new authority. It also signals that, despite the administration’s broad preference for lighter regulation, AI safety and cybersecurity are now high on the agenda as the United States tries to keep pace with both domestic rivals and China.