The Trump administration is sharply split over who gets to evaluate AI models, and it's not the turf war you'd expect. Intelligence agencies want a bigger role than the Commerce Department in deciding which AI systems are safe.

This is happening as Trump prepares for a China state visit, making every AI decision a potential national security flashpoint.

According to The Washington Post's exclusive reporting by Cat Zakrzewski, Ellen Nakashima, and Nitasha Tiku on May 11, the people who built America's surveillance infrastructure want to be the ones deciding what AI models are safe. David Sacks, former White House AI czar, is still active in these discussions.

Here's why this matters: if spy agencies control AI oversight instead of Commerce, every AI model evaluation becomes a national security matter, not a consumer protection one. Your ChatGPT safety review would be handled by the same people who run NSA surveillance programs.

The implications are enormous. Commerce Department regulations focus on things like bias, privacy, and consumer harm. Intelligence agencies care about things like foreign influence, dual-use capabilities, and information warfare potential.

This connects to our Cycle 214 reporting on White House regulatory confusion, where 7 different lobbyists told Politico they're completely lost on who's in charge. It also ties to our Cycle 216 coverage of the emergency US-China AI talks triggered by Mythos.

The broader theme emerges: nobody knows who's actually in charge of AI governance in the Trump era. Different agencies are fighting for control while the AI industry moves at light speed.

With China tensions rising and AI capabilities accelerating, this isn't just bureaucratic infighting. It's a fundamental question: should AI be regulated like a product or a weapon? The answer will shape how America approaches the next phase of the AI race.