A government cybersecurity advisory has revealed that several China-based AI companies are stealing artificial intelligence technology from American companies that pioneered the technology. Experts say the Trump administration must utilize commerce policies and sanctions to impose consequences on the Chinese entities involved in the theft.

On Tuesday, the FBI, the National Security Agency, and other federal agencies released a joint statement detailing how, since at least late 2024, Chinese AI companies like DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI have been fabricating AI models by “distilling” answers received from millions of queries directed at leading American frontier models, including variants of Claude, ChatGPT, Gemini, and Grok, “likely with Chinese government awareness.”

As noted by China experts Matt Pottinger and Liza Tobin, distillation can be done with permission for research purposes, but the Chinese companies are doing it “covertly and in breach of the terms of service” of American AI platforms by opening “tens of thousands of fake accounts” and asking the platforms “millions of questions,” recording the answers, and training their own models on them — thereby “inheriting much of the original’s capability without needing anything like the research and investment that produced it.” The result has been what Trump science policy adviser Michael Kratsios has called “industrial-scale” theft.

Xi Jinping’s regime is further utilizing this theft to censor information and spread malicious content and communist propaganda. A test of China’s DeepSeek model “found that it complied with 94 percent of overtly malicious requests (compared to eight percent for American models) and echoed four times as many false Communist Party narratives.”

Little has been done at a federal level to mitigate China’s massive theft of American technology over the last two decades, despite repeated denouncements from U.S. officials. The monetary losses due to the thievery are likely in the trillions of dollars, as the first Trump administration estimated at the time that the U.S. was losing anywhere from $250 to $600 billion annually.

Pottinger and Tobin argue that the Trump administration should take immediate action against the Chinese AI firms via the Commerce Department by barring their operation in the U.S. “The Commerce Department should restrict such models from American cloud platforms and block traffic to their developers’ APIs,” they contend.

Other experts like Lt. Colonel (Ret.) Robert Maginnis, who serves as senior fellow for National Security at Family Research Council, say that the implications of Beijing’s malicious activities are vast.

“The competition with China over artificial intelligence is no longer simply a commercial contest,” he told The Washington Stand. “AI has become an increasingly important element of national power, encompassing advanced semiconductors, data, computing infrastructure, energy, military applications, and the models themselves.”

Maginnis further observed that the latest findings of AI theft are “particularly troubling” due to the “scale and apparent organization of the activity. My position is that Washington must treat this kind of activity as a national security issue rather than merely a commercial intellectual-property dispute. That means protecting America’s most advanced models and semiconductor technology, strengthening counterintelligence and cybersecurity defenses, closing avenues used to evade technology restrictions, and imposing meaningful consequences on companies shown to have participated in illicit acquisition of American technology. The United States also needs to continue expanding domestic semiconductor production, energy generation, computing infrastructure, research capacity, and secure supply chains.”

At the same time, Maginnis argued, dialogue with the Xi Jinping regime is important as the Trump administration prepares to host the communist leader for a state visit later this month. “Diplomacy with Beijing remains necessary. The United States and China are major powers, nuclear competitors, and increasingly the world’s principal AI rivals. Maintaining channels of communication is therefore prudent. But diplomatic engagement should not mean that serious technology theft carries no consequences.”

“America cannot win this competition solely by restricting China,” he continued. “We have to out-innovate and out-build it. That requires strengthening what I describe as the American AI power stack: chips, electricity, data centers, research talent, manufacturing capacity, secure supply chains, and advanced models. We should also work closely with allies so that countries have a credible American-led technological alternative to dependence on Chinese systems.”

Maginnis further contended that “the larger strategic issue is straightforward: China should not be permitted to use access to America’s relatively open technological ecosystem to shortcut the enormous expense and effort required to develop capabilities that American companies spent billions of dollars creating. Washington should enforce American law, protect strategically important technology, and make clear that systematic acquisition of proprietary U.S. AI capabilities will carry consequences.”

“The danger otherwise is that American companies and taxpayers will finance much of the innovation defining the next era of economic and military power while Beijing systematically harvests the results,” he added.