Chinese military researchers are turning to leading US artificial intelligence models to help develop domestic defense AI, according to a Reuters investigation.
The report, based on more than 80 Chinese academic papers and patents, found that military-linked institutions have employed model distillation to build specialized systems for surveillance, cyber warfare, drone operations, and tactical decision-making.
The technique trains smaller AI models using outputs from larger ones, reducing computing requirements while allowing them to operate on resource-constrained military platforms such as drones, tactical vehicles, and surveillance systems.
The investigation identified models from OpenAI and Anthropic among those used in the effort.
Model Distillation Powers Defense AI
China has embraced model distillation to advance domestic military AI despite US export controls on advanced semiconductors, while accelerating the development of lightweight AI models and edge computing for battlefield platforms.
Sunny Cheung, a fellow at the Jamestown Foundation who reviewed more than 60 of the papers, said Chinese military scientists are systematically extracting reasoning from Western AI models to build indigenous defense systems.
“Teaching a model the right answer is one thing, but teaching it the reasoning behind the answer is much harder,” Cheung said.
“These papers show Chinese military-linked researchers are trying to transfer that expensive, proprietary reasoning from Western models into smaller systems they can control and deploy locally.”
One paper found that researchers from the People’s Liberation Army Unit 96941 used OpenAI’s GPT-3.5 to summarize sensitive military source code before training a domestic AI model capable of operating entirely within Chinese military networks.
Another described distilled AI models being used for image analysis on unmanned aerial vehicles and target recognition during simulated maritime operations involving drones, ships, and unmanned submarines.
Growing Scrutiny
The findings have intensified debate over the role of frontier AI models in military development.
Washington has questioned whether some Chinese organizations are using American AI systems to accelerate domestic capabilities.
Beijing rejected the allegations and accused the US of pursuing AI “hegemonism.”
Chinese AI developer Moonshot also denied claims that its Kimi K3 model relied on distillation from foreign AI systems, saying it was developed through proprietary technology.
