FBI breaks up Chinese AI smuggling network

U.S. officials have announced the dismantling of a major Chinese smuggling network that attempted to illegally export $160 million in advanced artificial intelligence (AI) chips to China.

The chips were subject to export control because of their advanced technology, said a Dec. 8 release by the Department of Justice (DOJ).

Law-enforcement agents took three men into custody, seizing over $50 million in high-performance graphics processing units (GPUs) plus cash.

Prosecutors called the case an important milestone as the first arrest and the first ever conviction for an AI diversion case.

“These chips are the building blocks of AI superiority and are integral to modern military application,” said U.S. Attorney Nicholas J. Ganjei. “The country that controls these chips will control AI technology; the country that controls AI technology will control the future.”

Alan Hao Hsu, who ran a Sugar Land, Texas, shell company called HAO Global, has pleaded guilty to smuggling charges as part of a widening federal case, reported Houston local media. 

Prosecutors say Hsu and two other businessmen helped funnel restricted high-end computer NVIDIA H100 and H200 chips into covert supply chains that ultimately sent the technology to China, despite U.S. export controls designed to stop exactly that. 

Hsu funded the scheme with more than $50 million in wire transfers from companies and individuals that tied back to China, including money from a logistics company based in Hong Kong, Ganjei said. 

Also charged in the operation are two men originally from the communist People’s Republic of China (PRC): Fanyue “Tom” Gong, 43, a PRC citizen living in Brooklyn, and Benlin Yuan, 58, a Canadian resident of Mississauga, Ontario. 

Gong was arrested Dec. 3 in New York, while Yuan was taken into custody Nov. 28 in Sterling, Virginia. 

Ganjei told a press conference in Houston the investigation, called Operation Gatekeeper, was underway for more than a year, noting the first arrests are “just the beginning of [the] operation.” 

The prosecutor added he’s confident other front companies are running the same smuggling racket. 

“The Southern District of Texas will find each and every person or business working to illegally export these critical technologies to our strategic competitors,” he added. “We will be relentless in finding them. That is what Operation Gatekeeper is for.” 

The FBI assistant director in charge of the New York field office, Christopher Reya, told the press conference that Gong allegedly hired people to relabel GPU shipments and disguise what was actually being exported. 

This enabled restricted U.S. technology to slip out of the country and into the hands of the Chinese government. 

Reya said the scheme amounted to a back-door pipeline to “provide an adverse nation with advanced technology using artificial intelligence applications and high performance computing to advance China’s civil military fusion efforts.” 

Millions of dollars of NVIDIA chips were held in various locations in New York in the scheme, where the labels were removed from the GPUs and relabeled under a fictitious corporate name to conceal the actual brand. 

“These events were witnessed by undercover law enforcement that had been planted at the warehouse,” Ganjei said. 

The shipping paperwork classified the goods as “adapters” instead of AI chips. 

A separate criminal complaint alleges Yuan helped recruit and organize inspectors to handle a batch of mislabeled GPUs for the Hong Kong logistics firm employing him, a company the DOJ describes as a Beijing front. 

Yuan allegedly instructed those inspectors not to disclose the chips were ultimately bound for China, the prosecutor said. 

The news came as it was reported by several media outlets that China’s AI startup DeepSeek, which rattled the AI market earlier in the year with low cost and high performance, are using illegally exported NVIDIA chips. 

Despite Beijing’s push to champion homegrown substitutes, Chinese-made AI chips still fall short of the performance required to train modern AI models, according to executives and engineers at Chinese AI firms, Investing.com reported. 

“AI, I think, has the potential to affect literally every scientific field, every field of study. It will touch upon economic concerns, military concerns,” Ganjei told reporters. “So, I would prefer that the United States be the leader in AI technology. I would strongly prefer that, and if somebody is illegally undermining that by sending these overseas, …. my office and (I) are very, very concerned about that.”