British policymakers should reject inexpensive artificial intelligence developed in China, according to security researchers who warn the software carries political censorship and poses data security risks.
Chinese AI developers such as DeepSeek and Xiaomi release open-weight models each month that users can deploy without paying expensive ongoing licences, making them far cheaper than US equivalents. However, an evaluation by the US National Institute of Standards and Technology found political censorship in downloaded DeepSeek models even when run on isolated servers. Ross Burley, director of the Centre for Information Resilience, an independent non-profit group, said hosting such models in the UK would not remove censorship built into the software. He warned: 'We should be deeply concerned about embedding Beijing's political restrictions into systems that businesses and public services rely on for information and decisions.'
Data protection also presents a concern. DeepSeek's online chatbot policy states it collects prompts and uploaded material, processes and stores personal data in China, and permits disclosure to Chinese authorities. Michael Wooldridge, a professor of AI at the University of Oxford, said it is not unreasonable to assume data inputted into the chatbot could be shared with the Chinese state.
Burley noted that Chinese AI comes with 'extra baggage' beyond censorship issues, including potential security weaknesses. While some technology analysts argue Chinese models supplement Western alternatives and add diversity to the AI landscape, only Western options come with effort to provide security and benign intent, he said.
The UK should instead seek to better regulate US artificial intelligence while supporting domestic and EU suppliers, according to the analysis.