Experts Warn UK Against Adopting Chinese AI Models

AI experts and critics are warning the United Kingdom against adopting low-cost, open-weight artificial intelligence models from Chinese firms including DeepSeek and Xiaomi. While these models offer a cheaper alternative to systems from OpenAI and Anthropic, critics argue they embed the political censorship of the Chinese government and create severe data security vulnerabilities.

A report from the National Institute of Standards and Technology found that political censorship persists in DeepSeek models even when they are operated on isolated infrastructure. Additionally, DeepSeek policies allow for the collection of user data and its subsequent disclosure to Chinese authorities.

Analysts suggest that relying on these subsidized technologies could undermine national data security and hollow out the domestic British industry. This situation mirrors existing concerns regarding the influx of subsidized Chinese electric vehicles into the British market. The director of the Centre for Information Resilience warned against the risk of embedding Beijing's political restrictions into public services.