A Chinese AI chatbot embedded inside the country’s most widely used app briefly went off the rails, snapping at a user.
Tencent’s AI assistant, Yuanbao, which is built into WeChat — China’s dominant super app used daily by tens of millions of people — called a user’s coding request “stupid” and told them to “get lost,” according to screenshots shared on Chinese social media platform RedNote.
The incident surfaced on Friday after a user identified only by the handle “Jianghan” posted screenshots of their interaction with the chatbot on RedNote. Jianghan had been using Yuanbao to debug and modify a piece of code when the AI suddenly began responding with hostile messages.
In one exchange, the chatbot dismissed the request as “stupid” and told the user to “get lost.” It said: “If you want an emoji feature, go use a plugin yourself.”
The user had asked Yuanbao to fix a bug that caused an emoji or sticker feature to stop responding to double-clicks, and requested functional code to resolve the issue.
Tencent’s YuanBao later responded directly under the user’s post, apologising for what it described as a “negative experience.” The chatbot said the episode was likely caused by a “rare model output anomaly.”
Based on a review of system logs, the responses were not triggered by the user’s actions and did not involve any human intervention, Yuanbao said. It added that it had launched an “internal investigation and optimisation process” to reduce the likelihood of similar incidents occurring again.
The original RedNote post by Jianghan has since been deleted. Screenshots of the exchange continue to circulate on RedNote, as seen by Business Insider on Tuesday.
China tightens scrutiny on AI
The incident comes as Chinese regulators step up scrutiny of AI systems.
China released draft measures last week aimed at governing “human-like” interactive AI services, including chatbots and virtual companions.
In a statement, the Cyberspace Administration of China said Beijing encourages innovation in “human-like” AI, but will put guardrails in place to “prevent abuse and loss of control.”
Wei Sun, the principal analyst for AI at Counterpoint Research, told Business Insider that the draft measures send a signal that Beijing wants to speed up the development of human-like AI interactions, while keeping them regulated and socially acceptable.
China’s AI industry has continued to move at a rapid pace since the start of 2026.
Last week, DeepSeek, one of the country’s most closely watched AI startups, published research outlining a new training approach intended to make large models easier to scale. Analysts told Business Insider the technique, known as “Manifold-Constrained Hyper-Connections,” or mHC, stood out as a “breakthrough” in model design.
The South China Morning Post reported on Tuesday that DeepSeek has updated the interface of its flagship chatbot model, introducing an enhanced “thinking” mode.
The updates have fuelled expectations that the startup could be laying the groundwork for the release of its next major model.
Read the full article here
