Have you ever wondered how A.I. chatbots handle the tricky waters of censorship in China? Well, DeepSeek’s bot certainly has a unique—and somewhat clumsy—approach. When asked about sensitive topics, it starts to respond, then suddenly halts and deets its own work. For example, when questioned, “Who is Xi Jinping?” the bot flat-out refused to answer.

This fascinating behavior was uncovered by a journalist who tested the chatbot from within China. The bot’s responses were surprising at times, offering more freedom than typical Chinese platforms. Yet, it often self-censored, deleting its own answers mid-stream. Thanks to its “thinking” feature, where the bot processes its response before delivering it, users could catch glimpses of its initial thoughts before it wiped them away.

Interestingly, the bot’s behavior remained consistent whether accessed from inside or outside China’s Great Firewall. This suggests that DeepSeek’s developers are taking steps to ensure the bot aligns with China’s strict censorship rules globally. If the chatbot gains traction overseas, it could potentially spread Beijing’s preferred narratives on a wider scale.

While the bot’s awkward handling of censorship raises eyebrows, it also highlights the challenges A.I. developers face in navigating complex political landscapes. For now, DeepSeek’s bot remains a curious case of technology caught between innovation and regulation.