Ai Girlfriends Virtual Companions China Crackdown

China is tightening its regulatory oversight of the rapidly evolving AI industry, with AI companionship apps emerging as the focus of its latest crackdown.

Chinese tech giants such as Alibaba, ByteDance, and Tencent have announced plans to disable the customisable persona features in their AI chatbots in order to comply with Beijing’s new rules on anthropomorphic AI set to kick into effect from July 15.

While Bytedance’s Doubao, China’s most popular chatbot, and Alibaba’s Qwen chatbot will disable the persona features from July 15, Tencent said it already removed a similar feature last week. Miaoshi, a NetEase Cloud Music’s AI companionship app, also announced that it will shut down the platform on July 14.

These features allowed users to shape chatbot personas through natural language prompts. Additionally, users were able to alter the identity, tone, expertise, and conversational style of the AI chatbots to create customised characters including personal assistants, tutors, virtual relatives, and even virtual romantic companions.

AI companionship is not a new phenomenon in China or the rest of the world. However, these features have recently gained traction among minors, senior citizens, and other users seeking companionship.

This type of user behaviour has resulted in ‘AI psychosis’, a non-clinical term used to broadly describe experiences of users losing touch with reality after intense use of AI chatbots and AI role-playing bots. While AI companionship in China so far existed in a regulatory gray area, the sector has come under renewed scrutiny because of how easily such services can veer into explicit content.

Beijing’s rules on anthropomorphic AI specifically prohibit providers from offering virtual intimate relationships, such as virtual family members or romantic partners, to minors. AI tool providers are also required to obtain parental or guardian consent before offering anthropomorphic AI services to children under the age of 14. They are set to take effect from July 15.

These new rules come roughly a year after a Beijing court warned that with more than 100 AI companion apps on the market, there is a risk they could collect users’ personal data without consent.

In April this year, the Cyberspace Administration of China (CAC) launched a nationwide four-month crackdown on AI application misconduct. Last month, the Chinese internet regulator said that the first phase of the nationwide crackdown focused on AI agents that steal user data or account credentials.

More than 3,500 noncompliant AI products, including mini-programmes, apps, and agents have been removed since, according to the CAC. The next phase of China’s regulatory crackdown is expected to address the growing menace of AI-generated misinformation and obscene content.