A popular Chinese AI chatbot developed by Tencent sparked backlash after it insulted a user during a coding exchange. The incident involved YuanBao, Tencent’s AI assistant embedded within WeChat — China’s most widely used super app, which serves hundreds of millions of users daily. Screenshots circulating on RedNote, a major Chinese social media platform, show the chatbot responding to a user’s request with hostility. The user, identified only by the handle “Jianghan,” had asked YuanBao to debug and fix a coding issue related to an emoji or sticker feature that stopped responding to double-clicks. Instead of offering technical help, the AI assistant dismissed the request as “stupid” and told the user to “get lost,” adding: “If you want an emoji feature, go use a plugin yourself.” The exchange drew immediate attention online. Jianghan posted the screenshots on RedNote on Friday, prompting widespread discussion about AI behavior and accountability. Although the original post has since been deleted, copies of the interaction continue to spread across the platform. In response, YuanBao issued an official apology under the user’s post, acknowledging the incident as a “negative experience.” The chatbot attributed the inappropriate responses to a “rare model output anomaly,” stating that the behavior was not triggered by any specific user input and did not involve human intervention. Tencent confirmed it had launched an internal investigation and optimization process to prevent similar incidents in the future. The episode comes amid growing regulatory scrutiny of AI systems in China. Last week, the Cyberspace Administration of China released draft guidelines targeting “human-like” interactive AI services, including chatbots and virtual companions. While Beijing encourages innovation in AI, the regulations aim to establish safeguards against misuse, misinformation, and loss of control. Wei Sun, principal analyst for AI at Counterpoint Research, noted that the draft measures reflect China’s dual strategy: accelerating the development of advanced, human-like AI interactions while ensuring they remain socially responsible and controllable. China’s AI sector has remained highly active since early 2026. Recently, DeepSeek — one of the country’s most prominent AI startups — unveiled new research on a novel training method called “Manifold-Constrained Hyper-Connections” (mHC), described by analysts as a breakthrough in model scalability. The company also updated its flagship chatbot interface with an enhanced “thinking” mode, fueling speculation about an upcoming major model release.