- Chinese tech giants disable custom AI personas by July 15.
- Regulations address explicit content and ‘AI psychosis’ concerns.
- New rules ban virtual intimate relationships for minors, require consent.
China is tightening its grip on the AI industry, with companionship apps becoming the latest target. Major tech companies, including Alibaba, ByteDance, and Tencent, have announced plans to disable customisable persona features in their AI chatbots to comply with Beijing’s new regulations on anthropomorphic AI, effective July 15. Miaoshi, NetEase Cloud Music’s AI companionship app, goes a step further, shutting down its platform entirely on July 14.
These persona tools have allowed users to craft AI characters ranging from tutors to virtual romantic partners.
Why China Is Regulating AI Companionship Apps
The persona features let users shape a chatbot’s identity, tone, and conversational style through simple text prompts. This created personalised AI characters, including virtual relatives, personal assistants, and romantic companions.
ALSO READ: Anthropic Found A Secret Thinking Space Inside Claude That You Never Get To See
While AI companionship has existed in China for some time, these features recently gained popularity among minors, senior citizens, and lonely users. Regulators grew concerned as such services often veered into explicit content and contributed to what some describe as “AI psychosis,” a non-clinical term for users losing touch with reality after prolonged AI interaction.
According to the Cyberspace Administration of China (CAC), more than 3,500 non-compliant AI products, including apps, mini-programmes, and agents, have been removed since the crackdown began in April. The first phase targeted AI agents stealing user data or account credentials.
What China’s New AI Rules Actually Require
Beijing’s rules specifically ban providers from offering virtual intimate relationships, such as romantic partners or virtual family members, to minors. Providers must also obtain parental or guardian consent before giving anthropomorphic AI access to children under 14. These rules take effect July 15.
ALSO READ: Zuckerberg Says AI Won’t Destroy Jobs, After Already Cutting 10,000 Of Them
The next phase of the crackdown is expected to address AI-generated misinformation and obscene content. A Beijing court had flagged data privacy risks from AI companion apps nearly a year ago, noting over 100 such apps were operating with little oversight at the time.
