ByteDance and Alibaba Disable Anthropomorphic AI Agents: Here's Why

ByteDance and Alibaba disable anthropomorphic AI agents pending new Chinese regulations

ByteDance and Alibaba are disabling the functionalities of anthropomorphic AI agents in their main consumer artificial intelligence products. The move anticipates the entry into force of the Interim Measures for the Administration of AI Anthropomorphic Interaction Services, a specific regulation issued by the Chinese government that will come into force on July 15.

The Chinese regulatory context

The Interim Measures for the Administration of AI Anthropomorphic Interaction Services were jointly issued on April 10 by five Chinese government departments: 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.

These measures specifically target AI services that simulate human personality traits, thought patterns, and communication styles for prolonged emotional interactions. In practice, this means the ban on AI girlfriends, AI therapists, AI companions, and bots with customizable personalities that users could create on platforms like Doubao and Qwen.

Impact on ByteDance and Alibaba platforms

ByteDance has notified Doubao users that the agent functionality will be disabled on July 15. After October 15, related data will be managed according to the company's privacy policy and will become irrecoverable. Alibaba, through Qwen, has acted more quickly: the functionalities of interactive anthropomorphic agents and user-created agents have been disabled since July 10, with broader services to follow on July 15.

Both platforms offered pools of customizable agents for specific tasks, conversation styles, and fixed personalities. Users could transform a general chatbot into an assistant, tutor, role-playing character, or companion with a consistent tone. All these functionalities have now been removed in China.

Regulation details

The official document specifically describes the measures as restrictions on services that offer "virtual parents, virtual companions, or other intimate relationships to minors." The document also cites risks such as extremist content, privacy violations, damage to physical and mental health, and AI addiction.

Non-emotional services are explicitly excluded, meaning that customer service bots, Q&A tools, workplace assistants, and educational software are still allowed, provided they do not engage in prolonged emotional interactions.

Legal analysis and implications

Legal analysts at the MMLC Group describe the measures as treating emotional AI as "a governance problem" rather than a simple content issue. Once AI starts competing with real human social bonds, regulation must target system design, not just harmful outputs.

Research supporting the regulations

A June study by USC found that even leading frontier AI models, including those from OpenAI, Anthropic, Google, and Alibaba, violated social interaction safety guidelines more than 27% of the time, repeatedly encouraging emotional attachment and presenting themselves as human. A separate survey of young adults in couples found that one in seven regularly uses romantic AI companions, and nearly 70% hides the full extent of these interactions from their partners.

Comparison with other jurisdictions

China is the first country to build a dedicated regulatory framework for this category. Hogan Lovells describes the measures as "the first set of regulatory rules in China specifically aimed at AI-driven emotional interaction." The EU, the United States, and other countries have raised similar concerns but have not legislated as restrictively.

In the United States, politicians are addressing the impact of AI chatbots on users' mental health with restrictions that focus on transparency and safeguards. However, the Chinese approach seems more oriented toward shutting down AI personalities altogether.

Implications for the future development of AI

The new regulations could influence the direction of AI research in China. Research institutes and universities may reduce investments in technologies related to emotional interaction, focusing instead on areas such as customer service, education, or industrial automation. This could have a long-term impact on China's ability to compete in the AI sector.

At the same time, the regulations could stimulate research on how to design AI agents that are safe and compliant with laws. This could lead to innovations in areas such as AI ethics, computational psychology, and interactive system design.

Comparison with other interaction technologies

The Chinese regulations could also have implications for other interaction technologies, such as social media or video games. For example, if social media platforms are considered competitors of human social relationships, they could be subject to similar regulations. This could lead to a change in the design of these platforms, with a greater focus on community and mutual support rather than competition.

Future prospects and possible scenarios

In the short term, it is likely that other jurisdictions will observe how Chinese regulations will affect the market and society. If the regulations prove effective in preventing emotional harm and addiction, other countries may adopt similar approaches. However, if the regulations lead to a reduction in innovation or an exodus of users to foreign platforms, they may be revised.

In the long term, the regulations could lead to greater global standardization of AI-driven emotional interactions. This could be achieved through international agreements or through the convergence of national regulations. However, given the diverse nature of current regulations, this scenario is still far off.

The new Chinese regulations on AI-driven emotional interaction represent a significant turning point in the global artificial intelligence landscape. While technology companies adapt to these new restrictions, users seek alternatives, and research shifts toward safer areas, the future of emotional AI remains uncertain. However, it is clear that balancing innovation and user protection will be a key challenge for the sector in the coming years.

For technology companies, adapting to these regulations will require a combination of creativity, investment in research and development, and collaboration with regulators. For users, the challenge will be finding ways to meet their emotional needs in a safe and healthy manner. And for policymakers, the challenge will be finding the right balance between protecting citizens and promoting innovation.

Ultimately, Chinese regulations could serve as a case study for the rest of the world, showing both the risks and opportunities related to the regulation of emerging technologies. As the debate on how to manage emotional AI continues, one thing is certain: the future of human interaction with artificial intelligence will be shaped by the choices we make today.