Chinese companies and researchers are using output from leading American AI models to train homegrown rivals, and the practice has moved from academic experiments into mainstream industrial use, according to a Reuters review of more than 300 Chinese documents published on November 17. Beijing is encouraging the practice, known as distillation of US AI models, as China works to offset tightening American restrictions on its access to advanced chips.

The documents include academic papers, patents, corporate disclosures, recruitment notices and government records. Together, they describe Chinese researchers using US models to produce answers, build code and solve difficult problems, with companies developing software to automate the process and workers preparing material for training smaller domestic models.

What Reuters found

The core finding is scale. Distillation, in which a powerful teacher model generates answers or demonstrates tasks so developers can train a smaller student system, is no longer a laboratory curiosity in China. Reuters found it embedded in industrial workflows.

The documents also show Beijing encouraging distillation to make AI cheaper and easier to deploy across China’s manufacturing sector. Government documents do not instruct companies to distil American models, but they encourage the technique as part of the push to spread AI across the world’s second largest economy. A national AI plus Manufacturing policy package published by eight government agencies in January includes guidance supporting model training, fine-tuning and distillation for industrial applications, and Shanghai and Shenzhen have encouraged techniques to reduce computing requirements.

Reuters also set out the limits of the technique. “Distillation is not illegal and is prevalent in other parts of the world,” the agency noted, adding that the process cannot reproduce a leading model’s full capabilities or eliminate the need for advanced computing power. What it can do is preserve selected skills in smaller, cheaper systems, potentially reducing reliance on American technology.

Beijing skyline with modern skyscrapers under a clear blue sky

Beijing. Government documents show China encouraging distillation as part of its push to spread AI across manufacturing. Image: N509FZ / Wikimedia Commons, CC BY-SA 4.0.

Three examples from the documents

The review cites concrete cases. In a November 2024 paper, Shanghai Jiao Tong University researchers described selecting about 5,000 answers from OpenAI’s o1 model, translating them with GPT-4o mini and assembling 10,750 bilingual training examples to further train Alibaba’s Qwen2.5-72B model. The researchers also cautioned that excessive reliance on distillation could stifle original innovation, leaving developers dependent on breakthroughs by more advanced model providers.

A researcher in a cleanroom holding a silicon wafer

A researcher holds a silicon wafer. Distillation cannot replace advanced computing power, but it can preserve selected skills in smaller, cheaper systems. Image: U.S. Department of Energy, public domain.

In an April 2025 paper, Alibaba researchers described a more systematic approach, using GPT-4 and GPT-4o to generate training material in languages other than Chinese, then refining and scoring responses before training smaller Qwen models. And in a February 2026 paper, researchers from Shanghai Jiao Tong University and food delivery company Meituan described using GPT-4o to generate plans for coordinating AI agents working together on programming problems, helping train a smaller model to organise such tasks.

Shanghai Jiao Tong University, Alibaba and Meituan did not respond to requests for comment. A former Alibaba engineer told Reuters, on condition of anonymity, that the attraction is making one expensive query do repeated work: an American model can answer a difficult question once, letting a Chinese team retain useful examples and train a local system around them.

Why Washington is worried

China’s broadening use of distillation could undercut US curbs on the sale of advanced chips to China, because its developers can build AI capabilities by drawing on their American rivals. Washington said on November 16 it would further restrict China’s access to such chips from companies like Nvidia, the day the Reuters review was published.

The issue is also a source of friction between the world’s two biggest economies ahead of the November 18 to 19 Asia-Pacific Economic Cooperation summit in Shenzhen, where Donald Trump and Xi Jinping are expected to meet. US authorities publicly accused Chinese companies in September of aggressive, targeted distillation of American AI models on an industrial scale.

At meetings in Chengdu in July, US Commerce Department officials raised concerns with Chinese regulators about developers collecting output from leading American AI models, warning some activities could amount to intellectual property theft, according to three people familiar with the discussions. Chinese officials defended distillation as a standard AI development technique and said they were open to acting where the conduct constituted IP theft, but made no concrete commitments, a US official familiar with the previously unreported exchange said. The US Commerce Department and China’s Cyberspace Administration did not respond to requests for comment.

The leading American AI companies already restrict the practice in their terms. OpenAI restricts using its output to develop competing models, subject to exceptions, while Anthropic similarly restricts training competing systems without permission. Anthropic has already cut off paying customers it caught distilling Claude.

Why it matters

A US official familiar with the July Chengdu discussions told Reuters that Washington believes industrial-scale distillation is the main reason Chinese AI models have narrowed the gap with their American rivals. Two people familiar with Chinese developers’ operations disputed that, saying distillation was only one factor alongside advances in engineering, algorithms, data and model design.

Either way, the review shows the dispute is now structural, not anecdotal. Distillation sits at the intersection of two trends: China’s need to build AI capability under chip curbs, and Beijing’s policy of pushing smaller, cheaper AI systems into factories. Concerned by the growing sensitivity, Chinese AI developers have in recent months asked Beijing’s industry regulator for guidance on the boundary between ordinary use of foreign models and organised collection of output for training, according to one person familiar with the discussions. China’s Ministry of Industry and Information Technology did not respond to a request for comment.

With the APEC summit opening on November 18, the findings land as both governments negotiate the terms of AI competition. The question is no longer whether distillation happens, but where the line sits between legitimate technique and industrial scale extraction.

Sources