

Instead of restricting American AI with bureaucracy, the president should focus on enabling rapid global deployment of American cutting-edge technology.
A mid the ever-ratcheting trade war, the Trump administration shocked the tech industry with something novel: AI trade restrictions will be eased. Just eight days before the so-called “AI diffusion rule” was set to enter force, the Department of Commerce announced it would be halted and soon replaced with a “much simpler rule.”
This choice is wise. The diffusion rule was more than just a mere export control. Its text, which rambled over hundreds of pages, constructed several global tiers of AI access and strictly dictated when, where, and how advanced AI products can be deployed. It was an overly ambitious bureaucratic mess.
As Deepseek and other Chinese AI companies challenge American technical leadership, worries that the Chinese Communist Party could wield AI to extend its surveillance reach or cement a military advantage have only grown. While the Department of Commerce has been clear that the rule in its current form is a no-go, it has stressed that concerns about China aren’t going away. Worryingly, Bloomberg reports the diffusion rule’s first-of-its kind export controls on advanced AI software may still be under consideration.
Proceeding on this or on any form of model weight controls would represent a grave threat to American AI success. If the administration were to implement the controls proposed under the diffusion rule, advance model exports would require a strictly vetted license. Even when exports are allowed, firms would be further required to tightly restrict AI product access. This applies even for exports to closely allied nations.
To deploy advanced models in either London or Vancouver, American firms would be required to lock down data centers with the same costly security standards that the military uses to guard top-secret intelligence. This is just one of countless controls.
The natural logic behind such rules is that perhaps through deep security rigor the government can stop American AI from inadvertently slipping into CCP hands. The problem here is that export controls rarely work as their planners imagine.
In the case of potential AI software controls, the risks are many. By imposing severe deployment restrictions and security costs on overseas models, the government risks capping how much American AI can be deployed abroad. Facing such limits, American AI companies may be incapable of providing the software bandwidth needed to meet international demand. This could dry up revenue streams necessary to finance future innovation and create market vacuums that Chinese alternatives would eagerly fill.
The biggest risks lie in technologies that don’t fit the bounds of the prescribed regulatory mold. Consider the AI-powered trucks, ships, or aircraft that could someday form the backbone of the transportation industry. The controls proposed in the diffusion rule make the mistake of demanding that advanced AI models remain restricted to military-grade data centers — a requirement that could exclude any advanced models designed to transport goods between nations. The hubris of such global regulation could kill off an American-led AI logistics industry even before its birth.
Instead of restricting American AI with bureaucracy, the president should focus on enabling rapid global deployment of American cutting-edge technology.
To best China, American models must be the global leaders on price, capability, and access. If our AI can readily jump from the lab and into use, we will ensure that American innovation remains the global default.
An essential first step is to avoid the AI model controls proposed by the diffusion rule. While more work to advance a competition-focused agenda is needed, this key step will help foster U.S.-led global prosperity.