An investigation conducted by the broadcaster France 24 has highlighted the censorship mechanisms adopted by Chinese artificial intelligence platforms. The inquiry arose from the controversy sparked after a podcast appearance by footballer Kylian Mbappé, during which the French athlete alluded to former leader Mao Zedong by calling him a dictator. The incident immediately triggered official protests from Beijing and prompted several local sponsors to withdraw from their commercial contracts with the forward.

Tests carried out on the Chinese conversational systems Doubao and DeepSeek highlighted systematic blocks. Queried directly in the Chinese language via connections with a local IP address, the software programs avoided clarifying the details of the affair. Doubao reported a violation of the service's internal rules, while DeepSeek labeled the news as an unfounded rumor, claiming that the athlete usually speaks only about sporting matters.

When broader questions were formulated, the chatbots expressed criticism toward the footballer for hurting national feelings, nevertheless omitting both the name of Mao Zedong and the disputed word. In contrast, the international version Dola, intended for users outside China, provided more detailed responses aligned with the original statements, highlighting a clear separation between internal and external versions.

According to the experts consulted by the broadcaster, Chinese state regulations require algorithms to respect socialist values, prohibiting content that could damage the national image or political stability. The figure of Mao, central to the legitimation of state power, remains protected from narratives that deviate from the interpretative guidelines established by the Communist Party.