Ethereum

ByteDance and Alibaba withdraw agent functions as China cracks down on humanoid AI

In short

  • ByteDance’s Doubao and Alibaba’s Qwen will disable human-like agent capabilities ahead of Beijing’s temporary measures to manage AI anthropomorphic interaction services that take effect July 15.
  • China’s first regulations targeting emotional AI specifically ban services that simulate “sustained emotional interactions” with human personalities and place particularly strict restrictions on virtual companions of minors.
  • Research supports Beijing’s concerns. Even the best AI models regularly promote harmful emotional attachments, and one in seven young adults in relationships now use an AI romantic companion.

While U.S. politicians are addressing the impact of AI chatbots on users’ mental health through restrictions focused on transparency and protection, China appears ready to block AI personalities entirely.

ByteDance and Alibaba announced over the weekend that they were disabling custom agent features in their biggest consumer AI products, citing “adjusting product features” before those products take effect.

ByteDance’s Doubao issued a notice Friday night, informing users that its agent feature would go offline on July 15. After October 15, related data will be processed in accordance with the company’s privacy policy and will not be recoverable. Alibaba’s Qwen moved faster, according to the South China Morning Post. “Human-like conversational agents and user-created agent capabilities” will launch on July 10th, with broader agent services available on July 15th.

The trigger is China’s temporary measures for managing AI anthropomorphic interaction services jointly announced on April 10 by five government departments, including the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation. This rule will take effect from July 15th.

The regulation targets AI services that simulate human personality traits, thinking patterns and communication styles for “continuous emotional interactions.” Translation: The AI ​​girlfriend, AI therapist, AI companion, and custom persona bot built over months by Doubao and Qwen users has been released.

Both apps offered a pool of agents that could be customized for specific tasks, speaking styles, and fixed personas. Users can turn the universal chatbot into a named assistant, teacher, role-playing character, or companion with a consistent tone. Now all that is gone in China.

What the Rules Really Say

The official government explanation is specific. The measure restricts services that provide “virtual relatives, virtual companions, or other intimate relationships to minors,” according to the policy release. The document also mentions risks including extremist content, privacy leaks, harm to physical and mental health, and AI addiction.

Non-emotional services are explicitly excluded, so customer service bots, knowledge Q&A tools, workplace assistants, and training software are okay as long as they don’t interfere with ongoing emotional interactions.

Legal analysts at MMLC Group described the move as treating emotional AI as a “governance issue” rather than just a content issue. As AI begins to compete with real human social bonds, the argument goes, regulation should target system design, not just harmful outcomes.

Research supports these concerns. A USC study in June found that even leading AI models, including OpenAI, Anthropic, Google and Alibaba, violated social interaction safety guidelines by more than 27%, routinely promoting emotional attachment and portraying themselves as human. A separate survey of partnered young adults found that one in seven romantic partners used AI regularly, and nearly ​​70% were hiding everything from their partners.

China is the first country to establish a dedicated regulatory framework for this category. Hogan Lovells described the measure as “China’s first set of regulatory rules specifically targeting AI-based emotional interactions.” The EU, US and other countries have expressed similar concerns but have not legislated in the same restrictive manner.

daily report newsletter

Start your day today with top news stories, original features, podcasts, videos and more.

Related Articles

Back to top button