Archive · 31 Jul 2026
China Uses US AI Models to Train Military Systems
Chinese scientists are taking the answers from big American AI programs like OpenAI’s GPT-3.5 and Anthropic’s Claude 3 Haiku and using them to teach smaller, cheaper AI models.
These smaller models can run on Chinese military equipment like drones and computers that don’t have huge power.
The goal is to get the same smart thinking without needing the expensive computers that the US uses.
The United States says this might steal technology and break export rules.
China says it is just learning from the best and that the US does similar things.
The whole thing is part of a bigger fight over who controls AI technology.
Chinese military researchers used outputs from OpenAI and Anthropic models to train domestic AI systems.
The technique, called model distillation, creates smaller models that can run locally without huge computing power.
The practice is widespread across Chinese defense institutions, including the PLA and national universities.
The U.S. accuses China of unauthorized extraction that could violate export controls and intellectual property rights.
China denies the accusations, claiming the U.S. engages in similar practices and that the models are used for national security.
- Who
- Chinese military researchers linked to the People’s Liberation Army and other defense institutions
- What
- Used outputs from US AI models to train smaller domestic AI systems
- When
- July 31, 2024
- Where
- Beijing, China
- Why
- To advance China’s defense capabilities and close the gap with US AI technology
- Technique
- Model distillation
- Key actors
- Chinese military researchers
- Source models
- OpenAI, Anthropic
- Purpose
- Train domestic AI for defense
- Controversy
- Unauthorized extraction allegations
