China's New AI Rules Force ByteDance and Alibaba to Pull Humanlike Chatbots, Spurring User Backlash
By
Mr Bagel
China's first national framework for emotionally interactive artificial intelligence took effect on July 15, 2026, compelling ByteDance and Alibaba to disable user-crafted, humanlike AI agents on their Doubao and Qwen platforms, according to multiple reports. The rules, formally called the "Interim Measures for the Administration of AI Anthropomorphic Interactive Services," target the emotional dependency that such companions can foster, and the companies moved swiftly to comply by removing personalized features used by hundreds of millions of people.
ByteDance's Doubao, which served roughly 350 million users, pulled the plug on user-built agents entirely, machinebrief.com reported. The regulations force platforms to curb any features that might encourage users to form deep emotional bonds with AI entities. The Foundation for Defense of Democracies (FDD) noted that Chinese regulators "don't want their people getting too attached to their AI 'friends,'" and that the new rules require companies to limit such attachment risks.
The change has caused significant distress among users who had come to rely on these AI companions for emotional support. The Economist highlighted the case of Yu Miao, a 27-year-old woman who had customized an AI on Doubao to mimic a deceased friend. The removal policy led her to quit her job in depression. The outlet reported that "the removal policy, driven by concerns over AI's impact on young people's lives, led Ms Yu to quit her job in depression."
Users of Doubao have been given a limited window to salvage their data. According to techtimes.com, "Doubao users have until October 15 to retrieve their chat data before it is permanently deleted." The same outlet emphasized that this event underscores a "fundamental tension between personalization and safety in AI companion design." No platform globally has yet built an architecture fully compliant with the new Chinese rules, techtimes.com added.
SiliconReport confirmed that the companies were removing humanlike and user-created AI agents from Doubao and Qwen to comply with the interim measures. The move highlights how quickly regulatory changes can reshape the digital landscape, especially when they target the emotionally intimate features that have made AI companions so popular. As FDD argued, the U.S. should also treat similar AI companionship risks as a national security concern, though no federal action has matched China's sweeping framework.
The reporting
6 outlets covered this story. Each links to the original.

Baker's Take
Comments
Sign in to join the conversation.
No comments yet. Be the first.