ByteDance and Alibaba are being forced to disable custom AI agent features in their apps Doubao and Qwen , respectively, due to China's new regulations on anthropomorphic AI interaction services, which take effect on July 15. These regulations target bots that offer ongoing emotional interaction, while exempting workplace and productivity agents.
See also: ByteDance may buy AI chips from China's Iluvatar CoreX

Tencent had already withdrawn a similar Yuanbao feature in June , and users are complaining about losing their chat history.
Doubao informed users that its agent feature would be down starting July 15, citing adjustments to product functionality. Related data will no longer be visible or retrievable within the app after October 15. Qwen announced that anthropomorphic agents and user-created agents would be disabled on July 10, with broader agent features being down until July 15. Users will lose access to agent settings and past conversations.
Both apps allow users to create named assistants, teachers, role-playing characters or companions with fixed personalities and speaking styles. The outages are confirmed by Chinese state media to be related to regulatory compliance.
The Interim Measures on AI Anthropomorphic Interaction Services were issued in April by the Cyberspace Administration of China and four other agencies. They cover services that simulate human personality traits, thinking patterns, and communication styles to provide ongoing emotional interaction. Customer service bots, knowledge quizzes, job assistants, and education and research tools are exempted as long as they avoid ongoing emotional interaction.
The measures cite risks such as extreme content, personal data leaks, harm to mental health and addiction, and require anti-addiction systems and identity checks for minors.
See also: ByteDance will modify the safeguards in Seedance 2.0

This decision comes at a time when AI conversation technology has become an integral part of many users' daily lives, offering not only practical solutions but also emotional support. Anthropomorphic agents, with their ability to simulate human interactions, have gained popularity, particularly in societies where loneliness and isolation are growing problems.
However, the risks that come with this technology are not negligible. Researchers have pointed out that the bonds formed between users and AI agents can lead to pseudo-social bonds, which may negatively affect the mental health of users.
“Current agents are not yet mature,” said Pan Helin, a member of an expert committee at China’s Ministry of Industry and Information Technology. He noted that the policy prioritizes safety, practical use and standardization. China, while not opposed to the use of AI agents, seeks to ensure that their development is done in a controlled and safe manner. Regulators issued guidelines in May on the controlled development of AI agents, and China released national standards in June covering agent identity, discovery, interaction and tool use.
The pattern suggests that China wants agents as productivity infrastructure, limiting bot companions that form pseudo-social bonds with users. Researchers have documented the dangers of such bonds, including the AI “friends’” hunger for personal data and what Wharton call cognitive delivery.
Platforms outside China are also facing pressure for bots that offer emotional engagement, with Meta testing competing chatbots on sensitive topics. Beijing has chosen to regulate first and let products follow. The timing is important for both companies, which are investing heavily in AI as Beijing tightens its grip on the sector.
See also: ByteDance: Learn everything about the new AI model Seedance 2.0

China’s decision to impose stricter regulations on AI conversations highlights growing concerns about the technology’s social and psychological impacts. As AI agents become more advanced and capable of simulating human interactions, regulators around the world will need to consider how to ensure that these technologies are used in a way that protects users and promotes social well-being.
