Donald Trump has mostly treated AI regulation as something to cut back. But on June 2, he signed an executive order called Promoting Advanced Artificial Intelligence Innovation and Security, the administration's first serious move toward frontier model oversight.
The core idea is a time advantage. Model developers may voluntarily provide newly trained frontier models to the government for 30 days before broader release, giving cyber defenders a chance to understand offensive and defensive capabilities, prepare patches and build detection tools. The order also sets deadlines: Treasury gets 30 days to build an AI cybersecurity clearinghouse, CISA must accelerate defensive deployment across civilian federal systems, and several agencies get 60 days to design classified benchmarks for AI-enabled cyberattack capability.
The key political line is that the order does not authorize mandatory licensing, pre-clearance or permits. That preserves the administration's pro-innovation posture while still acknowledging that advanced models may now matter for national cyber defense. In practice, the framework is voluntary on paper but hard for major labs to ignore if NSA decides a model belongs in the covered frontier category.
Sources：White House executive order text; NPR, Axios and Council on Foreign Relations analysis; CocoLoop