Chinese military researchers have used outputs from leading US AI models — including those from OpenAI and Anthropic — to train domestic systems for defense purposes, according to a Reuters review of more than 80 Chinese academic papers and patents, with research compiled by the Washington-based Jamestown Foundation.
The documents show widespread use of 'model distillation', a technique in which outputs from a powerful AI system are used to train smaller, specialized models that can run locally without the enormous computing requirements of building frontier AI from scratch. The papers suggest Chinese defense institutions see US models as both a source of technical insight and a shortcut to closing the gap with American rivals, despite Washington's export controls on advanced chips.
One paper by researchers in PLA Unit 96941, a Beijing military intelligence and cyber-warfare unit, described using OpenAI's GPT-3.5 to summarize sensitive military source code, then training a domestic model on those summaries so it could run entirely within Chinese military networks. Researchers at the North University of China, which has close ties to the weapons industry, used Anthropic's Claude 3 Haiku to generate synthetic training data for a social-media monitoring model. A 2024 paper from the PLA's National University of Defense Technology described distilling an image-processing model for deployment on drones, letting them analyze live video for navigation and targeting even when communications are cut.
The practice has become a flashpoint ahead of US-China talks on AI governance: Washington accuses some Chinese entities of extracting capabilities from American models, undermining export controls and infringing intellectual property. China rejects the accusation as AI 'hegemonism', and startup Moonshot last week denied US claims that its Kimi K3 model relied on distillation. 'Teaching a model the right answer is one thing, but teaching it the reasoning behind the answer is much harder,' said Jamestown fellow Sunny Cheung. '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.'
