US Intelligence Agencies Accuse Six Chinese AI Companies of Large-Scale Model Theft

U.S. intelligence and cybersecurity agencies have jointly issued an advisory alleging that six Chinese artificial‑intelligence firms are siphoning vast amounts of data from American AI models to fast‑track their own development. The notice, signed by the National Security Agency, the Cybersecurity and Infrastructure Security Agency and the Federal Bureau of Investigation, claims the companies have extracted billions of tokens—a measure of text used to train language models—without permission.

The firms named in the advisory include DeepSeek, Moonshot AI, Alibaba, MiniMax and two others that have not been publicly identified. According to the agencies, the stolen data encompasses not only raw text but also the sophisticated capabilities embedded in leading U.S. models, enabling the Chinese companies to replicate features that would otherwise require years of research and substantial computational resources.

U.S. officials say the activity raises serious concerns about intellectual‑property theft and national‑security risks, particularly as generative AI becomes integral to sectors ranging from defense to critical infrastructure. The advisory follows a series of warnings over the past year about foreign actors attempting to acquire advanced AI technology, and it underscores a growing focus on protecting the nation’s AI assets from covert extraction.

Washington’s response may include tighter export controls on AI‑related software and hardware, as well as increased scrutiny of cross‑border collaborations involving high‑performance models. Industry observers note that the United States has already begun to limit the export of certain AI chips and to require licensing for foreign entities seeking access to cutting‑edge tools, steps that could be expanded if the alleged theft continues unchecked.

The notice also serves as a signal to the broader tech community that the U.S. government is prepared to publicize and counteract illicit data‑harvesting efforts. While the advisory does not detail any immediate legal actions, it sets the stage for potential sanctions or other enforcement measures against the implicated companies. Analysts say the episode highlights the accelerating race for AI supremacy and the need for robust safeguards as the technology’s strategic value becomes increasingly evident worldwide.