US claims Chinese AI firms are carrying out ‘industrial-scale’ theft of trade secrets

Chinese artificial intelligence firms are committing “industrial-scale” theft of their American counterparts’ trade secrets to save money and bring the coveted technology to market faster, US federal agencies alleged Tuesday.

On Tuesday, the Justice Department said Chinese AI firms had stolen over 1.2 terabytes of confidential AI models and training data from American firms. The alleged theft involved companies such as SenseTime, Megvii and iFlytek, which are accused of accessing source code from Nvidia, Google and Microsoft between January 2022 and June 2023. Federal prosecutors estimate the illicit data would have saved the Chinese firms roughly $800 million in research and development costs. The indictments name six Chinese nationals and two corporate entities as co‑conspirators in the scheme.

U.S. tech firms have already begun allocating an additional $150 million annually to bolster intellectual‑property protection after the allegations. That spending increase is expected to raise the price of enterprise AI subscriptions by roughly 4 percent starting next quarter. For a typical business paying $20,000 a year for AI cloud services, the hike would add about $800 to the annual bill. Consumers who rely on those services for streaming recommendations or voice assistants could feel the impact through higher subscription fees or delayed updates.

The current allegations echo a 2018 Department of Justice report that identified China as responsible for over 57 percent of all economic espionage cases involving trade secrets. Since then, Washington has tightened export controls on advanced semiconductors, yet Chinese firms have increasingly turned to acquiring AI software through illicit channels. Insiders say the shift reflects a strategic pivot from hardware to software theft as chip sanctions make hardware acquisition harder. Analysts note that the pattern mirrors earlier waves of espionage in telecommunications and biotech, suggesting a broader playbook for catching up in critical technologies.

The Senate Intelligence Committee is scheduled to hold a hearing on March 20, 2024, to review the evidence and consider new sanctions on the implicated Chinese AI firms. A classified briefing for committee members is expected to be released on March 18, potentially revealing the exact volume of stolen data measured in petabytes rather than terabytes. If the briefing confirms larger losses, the administration may invoke the International Emergency Economic Powers Act to block certain AI‑related exports to China within 30 days. Counterintuitively, some analysts argue that the publicity could accelerate Chinese investment in open‑source AI projects, reducing reliance on stolen code.