

China is increasingly using artificial intelligence models developed by American companies to train its homegrown AI systems. Tightening US restrictions on advanced chips push Chinese developers to explore alternative ways to strengthen their technology.
A Reuters review of more than 300 Chinese academic papers, patents, corporate disclosures, recruitment notices and government documents found that the practice, known as distillation, has expanded beyond academic research into mainstream industrial applications.
The findings highlight how China is attempting to reduce its dependence on advanced computing hardware while continuing to develop competitive AI models.
Distillation involves a powerful AI model, known as a teacher model, generating answers and demonstrating how to complete tasks. Developers then use these outputs to train smaller student models.
Chinese researchers have used American AI systems to generate answers, write computer code and solve complex problems. Companies are also developing software to automate the process, while specialist workers prepare training material for smaller domestic models.
In a November 2024 paper, researchers from Shanghai Jiao Tong University described selecting around 5,000 answers from OpenAI's o1 model, translating them using GPT-4o mini and creating 10,750 bilingual training examples to further train Alibaba's Qwen2.5-72B model.
Another study published in April 2025 detailed how Alibaba researchers used GPT-4 and GPT-4o to generate multilingual training material for smaller Qwen models.
Also Read: OpenAI Flags 16,000 ChatGPT Extraction Attempts, Accuses Moonshot AI
The growing use of distillation has become a point of disagreement between Washington and Beijing. US officials have warned that collecting outputs from American AI models on an industrial scale could cross into intellectual property theft.
During meetings in Chengdu in July, US Commerce Department officials raised concerns with Chinese regulators. Chinese officials defended distillation as a standard AI development technique but made no concrete commitments to address Washington's concerns.
OpenAI's terms restrict using its outputs to develop competing models, subject to exceptions. Anthropic also restricts training competing systems without permission.
China's government has encouraged distillation to make AI cheaper and easier to deploy across manufacturing industries.
However, researchers caution that excessive dependence on foreign models could limit original innovation. They also stress that distillation cannot replicate a leading model's full capabilities or eliminate the need for advanced computing power.