A well-meaning regulatory effort in Congress intended to ensure American AI dominance over competing Chinese models may have the opposite effect.

When Japanese startup Sakana AI recently launched its Fugu Ultra model, it sent a message to American policymakers: unpredictable access to American AI products creates market opportunities that foreign competitors are eager to fill.

Fugu Ultra is built to help customers switch among AI systems when one becomes unavailable. The timing of its release, coming shortly after US export controls disrupted access to Anthropic’s top-tier Fable and Mythos models, was hardly coincidental. Sakana seems to be betting that American AI will become less reliable over time, leading its current customers to seek better models further afield.

American Leadership on AI Isn’t Inevitable—Tread Carefully

That bet reflects something important about what’s actually at stake. Americans have grown accustomed to the fruits of US technological leadership. The internet, smartphones, and cloud computing transformed how the world works, and the productivity gains, innovation, and economic dynamism they generated flowed directly from American companies becoming the world’s default platforms. 

It has become almost cliché to complain about the faults of Big Tech, and sometimes those complaints have merit. But if leading technological platforms had been built in Beijing rather than Silicon Valley, Americans would have far more to complain about. US technological dominance was not an accident. It was the return on decades of American leadership in building technology the rest of the world chose to adopt.

AI represents the next technological inflection point, and the stakes may be even higher. The countries and companies that establish themselves as foundational AI providers—whose models become critical infrastructure for businesses and governments worldwide—will shape the trajectory of technological innovation for a generation. This is an enormous opportunity, and Washington risks squandering it through poorly calibrated regulation.

How Much AI Regulation Is Too Much?

Of course, export controls on AI do present a genuine dilemma. No one seriously disputes that keeping advanced AI out of hostile military powers is a legitimate goal. The problem is that poorly calibrated restrictions don’t just inconvenience adversaries. Instead, they also push allies and legitimate users toward alternatives, including increasingly capable Chinese models.

This matters more than it might seem. Switching AI models incurs substantial costs. Companies and governments that structure operations around a given platform tend to stick with it. The providers that win users today are likely to keep them for years. Reliability, not just capability, often drives those decisions. And right now, Washington may be making American AI less reliable.

Pending legislation like the Remote Access Security Act is likely to compound the problem. As mentioned, proponents have legitimate reasons to worry about adversaries accessing American computing infrastructure remotely. But as currently written, RASA’s scope is broad enough to burden allies and commercial users who pose no meaningful risks, leaving them little choice but to build their own solutions or turn to non-American AI providers.

Perhaps the strongest counterargument is that some short-term disruption is an acceptable price for keeping transformative technology away from adversaries. In practice, however, restrictions that are broad, unpredictable, or poorly targeted may do more to accelerate the development of non-American alternatives than to eliminate genuine threats. That, in turns, shrinks the global footprint of US technology while weakening America’s long-term strategic position in the process.

The lesson of the past few decades is that American technological leadership delivers returns that are hard to appreciate until you imagine their absence. Export control policy should be narrow, predictable, and focused on real threats. Treating disruption abroad as an acceptable byproduct, or as someone else’s problem, is a mistake the United States cannot afford as the foundational decisions about AI adoption are being made right now.

About the Author: Clark Asay

Clark D. Asay is Terry Crapo Professor of Law at Brigham Young University. Before he came  to BYU, Professor Asay was a visiting assistant professor and Shughart Scholar at Penn State’s Dickinson School of Law from 2012 through 2014. Prior to entering legal academia, Professor Asay worked at Amazon’s Lab126 and supported the Kindle, Kindle Fire, and Amazon Fire teams. His research and teaching interests focus on the intersection of law, technology, and innovation.

The post Export Controls and the AI Race: When Security Policy Undermines Security appeared first on The National Interest.