The Remote Access Security Act should stop China—not make American AI infrastructure harder for allies to access.

DeepSeek, a Chinese startup, just kicked off an AI arms race, announcing it intends to build a machine that thinks like a person, which the field calls artificial general intelligence (AGI). The announcement was also a recruiting call: the company plans to double in size, and more than 10,000 resumes flooded its inbox within a day. DeepSeek is hiring like a company whose government treats artificial intelligence (AI) as a national security imperative. The United States should treat it as one of ours, and to do so, Congress must ensure our allies build on American infrastructure rather than on technology subsidized by Beijing. 

DeepSeek was ostensibly built with private money, funded for years through founder Liang Wenfeng’s hedge fund. But in China, the line between a private company and a state one is practically non-existent: the 2017 National Intelligence Law compels every firm to support and cooperate with state intelligence work, and from the perspective of US regulators, Wenfeng’s hedge fund and DeepSeek are directly connected to the Chinese state apparatus. In DeepSeek’s first outside raise this year, China’s state-backed National AI Industry Investment Fund took the only voting rights in the deal, while commercial investors like Tencent got none. Security researchers have tied DeepSeek’s web login code to China Mobile, the state-owned telecom that American regulators barred from operating in the United States, and federal officials have weighed placing the company on the Commerce Department’s Entity List. 

None of this is about chat-based AI. Beijing wants something more powerful: an artificial intelligence system running its factories, laboratories, and weapons, tightening its grip on every economy it can reach, and it has said so on the record. China’s 2017 strategy set a deadline to lead the world in AI by 2030, and its newer “AI Plus” plan aims to embed the technology across sectors, from logistics and healthcare to finance and government, with ambitions to build a fully AI-integrated economy by 2035. China set a date; America has yet to answer with its own. 

Congress Must Strengthen America’s AI Infrastructure and Global Competitiveness

The federal government, to its credit, treats AI leadership as a national priority. The Stargate initiative aims to put up to $500 billion in private capital behind American data centers built to keep the country ahead of China. And this fall, the Export-Import Bank launched the American AI Exports Program to finance overseas purchases of trusted American AI so that allied governments build on our technology instead of on systems the Chinese state subsidizes. This is a key approach to ensuring American AI dominance, as every foreign customer on the American stack is one that is not on Beijing’s. 

The subtlest threat to that strategy comes from Washington itself. 

Congress is weighing the Remote Access Security Act. Its sponsors want to stop China from renting American computing power through the cloud to get around the export controls on advanced chips. Their objective is a worthwhile one. But the legislation sweeps too broadly and imposes restrictions on allied and adversarial nations alike, potentially discouraging legitimate firms in allied countries from building on American AI infrastructure. 

The bill hands the executive branch a broad new licensing authority over who may access American computing power from abroad. The Entity List, by contrast, names a specific bad actor after a finding, and its restrictions fall on that one company. This new authority is aimed at no one in particular. It is available to whoever runs the Commerce Department, and the administration that first wields it will not be the last. An administration concerned more about ideology than national security could turn the same power against allied companies and American cloud providers. That’s a risk that companies in friendly nations will take note of when considering their technology partners. At the very moment the United States should be expanding its global AI market share, Congress should be careful not to make American cloud services more difficult to buy than China’s. 

Every AI policy should be judged by one question: does it strengthen America’s position or China’s? Better models alone will not win this. The country that wins will be the one the rest of the world still chooses to build on, invest in, and buy from. 

China has shown how badly it wants to win the AI race, and it has the resumes to prove it. America still holds the stronger hand: the deeper capital markets, the better universities, the wider network of allies, and the technology those allies actually want to buy. Those are enormous advantages. The fastest way to lose them is to pass a law so broad that America’s own friends decide our AI is more trouble than it is worth. 

About the Author: Gregory S. McNeal

Gregory S. McNeal, JD/PhD, is a professor of Law and Public Policy at Pepperdine University and a nationally recognized scholar of law and technology. His research explores the legal and institutional frameworks that govern emerging technologies. He holds an MPA from American University, a JD from Case Western Reserve, and a PhD from Pennsylvania State University. 

The post China Is Recruiting for the AI Race. America Must Respond.  appeared first on The National Interest.