The move comes after months of internal discussions within the administration over the implications of AI for national security and cyber defence. In a reversal of its earlier light-touch stance, the White House has now opted for greater oversight of rapidly advancing AI systems.
The latest directive follows the withdrawal of a previous executive order last month that would have established a review period ranging from 14 to 90 days for new AI models. That proposal was abandoned shortly before it was due to be signed by Trump.
Under the new framework, companies developing advanced AI systems will be expected to submit their models for government assessment 30 days before public release. The administration said the shorter review period is intended to balance oversight requirements with continued technological progress.
The order also instructs the Treasury Secretary to establish an AI cybersecurity clearinghouse tasked with examining security weaknesses identified by AI models and coordinating responses to potential risks.
According to the executive order, while cutting-edge AI capabilities can strengthen the United States, they also create national security challenges that demand coordinated action across government departments and agencies.
The decision signals a departure from the administration’s previous position, which emphasised minimal regulation and argued that fewer restrictions would help the United States maintain an advantage in the technological competition with China.