U.S. Accuses Chinese Firms of Illicit AI Model Distillation

U.S. cyber and law enforcement agencies have accused Chinese AI companies of copying American AI models through distillation, a cost-lowering technique. Firms like DeepSeek and Alibaba allegedly used American models with possible government backing. These accusations arise ahead of U.S.-China talks on AI safety risks.

American security agencies have accused several Chinese AI companies, including Alibaba and DeepSeek, of illegally copying sophisticated U.S. AI models. These activities reportedly involve distillation, a method to train smaller AI models using larger ones. The accusations come from the National Security Agency, the Cybersecurity and Infrastructure Security Agency, and the FBI.

The agencies claim these distillation activities occur on an industrial scale, with potential awareness from the Chinese government. This issue surfaces as the U.S. prepares for Chinese President Xi Jinping's upcoming visit and a planned dialogue on AI safety risks between the nations.

The allegations highlight ongoing tensions between the U.S. and China regarding technology theft and AI safety, with both countries intent on addressing these critical topics in upcoming negotiations.