US Accuses Chinese AI Developers of Industrial-Scale Technology Theft
The US government has formally accused prominent Chinese AI developers, including DeepSeek and Alibaba, of malicious copying and industrial-scale theft of American AI technology. The accusations focus on unauthorized model distillation techniques used to replicate capabilities of frontier American models. This development intensifies the global AI race and technological friction between the US and China. It could trigger stricter legal regulations, tighter sanctions on open-weights model distribution, and enhanced API security measures across the AI ecosystem. US officials highlight the unauthorized use of proprietary AI outputs to train competing open-weight models like DeepSeek-R1 at a fraction of original development costs. US lawmakers are responding with legislative initiatives, such as the Deterring American AI Model Theft Act, to curb foreign distillation campaigns.
## BACKGROUND
Model distillation is an AI training technique where a smaller 'student' model learns by imitating the outputs of a larger, more complex 'teacher' model. While distillation is standard practice for creating efficient models, using proprietary frontier models as teacher sources without consent has sparked fierce debates over intellectual property rights.