China is set to tighten its grip on artificial intelligence with a new set of regulations targeting AI companionship platforms and customisable chatbot personas. Beginning July 15, several of the country’s leading technology companies will either remove or significantly limit features that allow users to create virtual companions, including AI girlfriends and AI family members, in an effort to comply with Beijing’s latest rules on anthropomorphic AI.
Major Chinese technology firms, including Alibaba, ByteDance, and Tencent, have already announced changes to their AI chatbot services. ByteDance’s popular chatbot Doubao and Alibaba’s Qwen will disable persona customisation features from July 15, while Tencent confirmed it had already removed similar functionality. Meanwhile, Miaoshi, an AI companionship platform operated by NetEase Cloud Music, will shut down entirely on July 14, just a day before the regulations come into force.
If you enjoyed reading this article, you can help support our journalism by adding OHeraldo as a preferred source.
The affected features previously allowed users to customise AI personalities using natural language prompts. People could modify a chatbot’s identity, tone, expertise and conversational style, transforming it into a virtual tutor, assistant, family member or even a romantic companion. These capabilities have become increasingly popular among users looking for emotional support and personalised interactions.
Although AI companionship services have existed in China and other countries for several years, their rapid adoption has drawn increasing attention from regulators. The technology has reportedly become especially popular among teenagers, elderly users and individuals seeking companionship, raising concerns over emotional dependency and online safety.
Authorities are also worried about what has been informally described as “AI psychosis”—a term used to describe situations where prolonged engagement with AI chatbots or role-playing applications may blur the line between reality and virtual interactions. While the phrase is not recognised as an official medical diagnosis, it has become part of the broader debate surrounding the psychological impact of AI companionship.
Another major concern for regulators is the possibility that AI companion services could drift into sexually explicit or otherwise inappropriate content. Until now, many of these platforms operated in a legal grey area, but the new framework seeks to establish clearer boundaries.
Under the regulations taking effect on July 15, AI providers are prohibited from offering virtual intimate relationships—including AI romantic partners or virtual family members—to minors. Companies must also obtain parental or guardian consent before providing anthropomorphic AI services to children under the age of 14.
The latest measures build on China’s broader effort to regulate artificial intelligence. Last year, a Beijing court warned that the country’s growing AI companion market—estimated to include more than 100 applications—posed risks related to the collection and misuse of users’ personal data without proper consent.
China has also intensified its wider crackdown on AI misuse. In April, the Cyberspace Administration of China (CAC) launched a nationwide four-month campaign targeting misconduct involving AI applications. According to the regulator, the first phase focused on AI tools that stole user data or account credentials, resulting in the removal of more than 3,500 non-compliant AI apps, mini-programmes and AI agents.
Officials have indicated that the next phase of enforcement will target AI-generated misinformation, deepfakes and obscene content, signalling that China intends to maintain strict oversight as generative AI technologies continue to evolve. The latest restrictions on AI companions reflect Beijing’s broader strategy of balancing technological innovation with tighter regulation over digital platforms and online behaviour.
