China’s top internet regulator has unveiled draft rules aimed at protecting minors from potential harms linked to artificial intelligence-powered services on social platforms and gaming sites. The proposed regulations seek to restrict the offering of “virtual intimacy services” to individuals under 18 and impose tighter controls on AI-based interactions that could negatively affect young users.
Published on Friday by the Cyberspace Administration of China (CAC), the draft titled “Ensuring minors’ safe and healthy use of the internet” outlines new obligations for online platforms. Among the key provisions, the rules would prohibit social networking services among strangers for minors under 16, except for those who are over 16 and earn their own income. The regulations also ban platforms from providing “virtual relatives or companions” that simulate real family or partner relationships targeting those under the age of 18.
Operators of online games, social media, and AI services would be required to activate special “minors’ mode” features designed to limit exposure to content that could harm cognitive development or mental health. Platforms are expected to improve methods for accurately verifying users’ ages and to refrain from deploying algorithms that encourage addiction, emotional dependence, or excessive spending among children and teenagers.
Penalties for violations could be substantial. If a platform’s revenue exceeds 1 million yuan (approximately HK$1.7 million), regulators could impose fines of up to 10 times the income earned, as well as confiscate all illicit gains.
Experts note that defining what constitutes a “virtual intimate relationship” is central to enforcement. He Qianggao, a lawyer and director at Geason Law Offices, explained that services specifically designed to simulate or replace real-world familial or romantic bonds, using character settings and emotional guidance, fall under scrutiny. In contrast, routine AI-powered chat functions or general role-playing are less likely to be targeted unless they foster emotional dependence that substitutes for human interaction.
He also noted that the challenges for AI companies go beyond simply verifying users’ ages. Once identified as minors, platforms must carefully manage what types of services can be accessed, potentially increasing costs related to product design, technical infrastructure, and compliance efforts.
The draft follows earlier measures introduced in July, which established China’s first dedicated rules for humanlike AI interaction. Those rules targeted AI systems that excessively pander to users or encourage emotional dependence to the detriment of real-world relationships. In response, major technology firms such as ByteDance and Alibaba Group Holding have disabled certain AI companion functions previously available on their platforms. For instance, Alibaba’s Qwen and ByteDance’s Doubao no longer support customized AI agents with fixed personalities or names.
The CAC has opened the draft regulations for public comment until October 17, seeking input from stakeholders ahead of finalizing the new framework to safeguard minors’ well-being online.