ByteDance and Alibaba Disable AI Agent Features Ahead of China's New Humanlike AI Regulations
Key Takeaways
- ▸ByteDance's Doubao and Alibaba's Qwen are disabling their customized AI agent features by July 15 in compliance with new Chinese government regulations
- ▸China's new "Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interaction Services" targets AI services that simulate human personality traits for sustained emotional interaction
- ▸The regulations exempt certain AI applications like customer service bots and educational tools that do not involve sustained emotional interaction
Summary
ByteDance's Doubao and Alibaba's Qwen are disabling their customized AI agent features in advance of China's new "Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interaction Services," set to take effect on July 15, 2026. Both apps informed users that features allowing the creation and customization of agents with distinct personas, skills, and communication styles will be taken offline. Doubao will disable its agent feature on July 15, with related user data to be handled according to privacy policies and become inaccessible after October 15. Qwen will begin disabling humanlike interactive and user-created agent functions on July 10, with broader agent services going offline on July 15.
The timing of these moves reflects Beijing's broader effort to regulate the fast-growing AI sector. The new measures specifically target AI services that "simulate human personality traits, thinking patterns and communication styles to provide sustained emotional interaction." The regulations notably exempt certain applications such as customer service bots, knowledge Q&A systems, workplace assistants, and education and research tools—provided they do not involve sustained emotional interaction. The rules cite significant concerns including the spread of extremist ideas, privacy leaks, potential harm to physical and mental health, and the risk of user dependency or addiction.
- The rules highlight government concerns about AI-related harms including extremism, privacy breaches, mental health impacts, and user dependency


