Story
September 10, 2026

US Warns China’s AI Push Is Built on Copying—Beijing Calls It a Smear

Washington portrays the alleged model-copying campaign as a direct threat to its technological lead, while Beijing says its AI gains reflect domestic innovation and cooperation—not theft—and casts the accusations as political smears.

The dispute sharpened after U.S. AI companies spent the past year warning that competitors were using model outputs to accelerate their own development. Distillation itself can be a legitimate machine-learning technique, but American officials say the alleged Chinese activity crossed into systematic extraction of proprietary capabilities.

On Sept. 8, the NSA, CISA and FBI named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI in a joint advisory. The agencies alleged that, since at least late 2024, the companies had used huge volumes of requests, fraudulent accounts, proxies and prompt-injection tactics to pull capabilities from Claude, GPT, Gemini and Grok. The activity was "likely" carried out with Chinese government awareness, though the advisory did not say Beijing directed individual campaigns.

Washington’s case is economic as much as technical: agencies said industrial-scale distillation can cut development timelines and training costs, allowing rivals to obtain frontier functions without shouldering the full expense of building them. They urged providers to tighten identity checks, spot coordinated accounts and share intelligence—measures that could also degrade service for legitimate users caught in the dragnet.

Beijing answered the following day with a flat rejection. Foreign Ministry spokeswoman Mao Ning said China’s AI advances stem from domestic technological development and international cooperation, while calling on U.S. agencies to pursue cooperation rather than make "groundless accusations." A Chinese Embassy spokesperson had earlier described the U.S. campaign as a "smear" driven by prejudice against Chinese technological achievements.

The clash fed directly into the broader race. Treasury Secretary Scott Bessent argued that U.S. leadership depends on continuing to build AI infrastructure, especially data centers: "If they were to pull ahead of us on AI, then nothing else matters." Beijing sees an effort to discredit its progress; Washington sees a warning that the lead it holds may not last if it fails to defend it.