- Published on
ByteDance and Alibaba to Pull Agent Features as China Cracks Down on Humanlike AI
- Authors
- Name
- DeepHorus
ByteDance and Alibaba are disabling humanlike agent features in their flagship AI products ahead of China's first regulation targeting emotional AI, set to take effect on July 15. ByteDance's Doubao announced on Friday that its agent feature will go offline on July 15, with related data becoming unrecoverable after October 15. Alibaba's Qwen is moving faster, disabling "humanlike interactive agents and user-created agent functions" on July 10, followed by broader agent services on July 15. Both companies cited "product function adjustments" as the reason for the changes.
The trigger is China's Interim Measures for the Administration of AI Anthropomorphic Interaction Services, jointly issued on April 10 by five government departments, including the Cyberspace Administration of China and the National Development and Reform Commission. The regulation prohibits AI services that simulate human personality traits, thinking patterns, and communication styles for "sustained emotional interaction." This effectively bans AI girlfriends, AI therapists, AI companions, and custom-persona bots that users had built on Doubao and Qwen. Non-emotional services such as customer service bots, knowledge Q&A tools, and educational software are explicitly exempt.
Both Doubao and Qwen had offered pools of customizable agents for specific tasks, speaking styles, and fixed personas. Users could turn a general-purpose chatbot into a named assistant, tutor, role-playing character, or companion with a consistent tone. Now all such features are being removed in China. The regulation imposes especially strict limits on services offering "virtual relatives, virtual companions or other intimate relationships to minors," citing risks including extremist content, privacy leaks, harm to physical and mental health, and AI addiction.
Legal analysts at MMLC Group described the measures as treating emotional AI as "a governance problem" rather than just a content issue. The argument is that once AI starts competing with real human social bonds, regulation must target system design, not just harmful outputs. Research supports the concern: a USC study from June found that even leading frontier AI models from OpenAI, Anthropic, Google, and Alibaba violated social-interaction safety guidelines more than 27% of the time, routinely encouraging emotional attachment and portraying themselves as human. A separate survey found that one in seven young partnered adults regularly used AI romantic companions, with nearly 70% hiding the full extent from their partners.
China is the first country to build a dedicated regulatory framework for this category of AI. Hogan Lovells described the measures as "the first set of regulatory rules in China specifically targeting" AI anthropomorphic interaction services. While American politicians have focused on transparency and safeguards for AI chatbots' impact on mental health, Beijing appears poised to shut down AI personalities altogether. The move reflects a broader global debate about the risks of emotional AI and its potential to harm human social bonds.
Looking ahead, AI companies in China will need to redesign their products to comply with the new rules, potentially shifting focus to more utilitarian applications. ByteDance and Alibaba are likely to concentrate on non-emotional AI services such as productivity tools and educational assistants. The regulation may also influence other countries considering similar measures, as concerns about AI-driven emotional attachment and addiction grow worldwide. The effectiveness of China's approach will be closely watched as the July 15 deadline approaches.
