Reuters: Chinese military researchers use OpenAI and Anthropic AI technologies

Chinese military researchers are using advanced artificial intelligence technologies developed by US companies OpenAI and Anthropic to build domestic AI systems for military applications, according to a Reuters investigation.

The report, produced in cooperation with the Jamestown Foundation, examined more than 80 Chinese academic papers and patent filings and found that institutions linked to the People’s Liberation Army (PLA) have been exploring ways to adapt foreign AI capabilities for defence-related purposes.

Researchers are reportedly using a technique known as “model distillation”, which allows smaller AI models to learn from the outputs and reasoning patterns of more powerful systems. The approach enables the creation of compact, specialised AI tools that can operate locally without relying on cloud-based services or access to advanced US semiconductor technology.

Sunny Cheung, a researcher at the Jamestown Foundation, said Chinese specialists are effectively transferring capabilities from leading Western AI models into smaller systems that can be independently controlled and deployed within China.

According to Reuters, PLA-linked research has explored the use of such technologies in areas including cybersecurity, source code analysis, drone operations, image processing, social media monitoring, and target identification for naval missions.

Anthropic said it does not provide commercial access to its products in China and actively monitors possible violations of its usage policies. The company has previously restricted access to its services in certain regions due to concerns over misuse.

The findings have raised concerns in Washington over whether the use of foreign AI technologies by Chinese military institutions could weaken US export controls and create intellectual property risks. Beijing has rejected such accusations.