Is China Stealing American AI? Why ‘Distillation’ Has Washington Up in Arms

U.S. security agencies have publicly accused six Chinese AI firms of systematically using American models to train their own systems, a technique commonly described as distillation. The claim escalates the U.S.-China AI rivalry from a chip-export fight into a dispute over the models themselves, and raises questions about how leading labs protect their outputs from being copied via API access.

• U.S. security agencies named six Chinese AI companies as allegedly exploiting American models for training.
• Distillation—training a model on another model's outputs—is now a national security flashpoint, not just a technical shortcut.
• Expect pressure on U.S. AI firms and policymakers to tighten access controls and terms of use around frontier models.