Chinese regulator drafts rules to protect teenagers from ‘intimate’ AI companions

China's Cyberspace Administration has proposed new regulations to safeguard young users from "intimate" AI companions, which could gain popularity as technology advances. The draft, published on Friday, would prohibit platforms from offering virtual relationships or companionship services to those under 18, as well as any features designed to compel minors into addiction or negatively impact their well-being.

Additionally, the rules would ban strangers from engaging in online social interactions unless they are aged 16 or above and earning their own income. Operators would be required to identify underage users more accurately and restrict access to algorithms that could lead to emotional reliance, addiction, or excessive spending. Penalties for non-compliance could reach up to ten times the income generated by the platform if it exceeds 1 million yuan (approximately US$149,000).

These proposals, currently open for public comment until October 17, reflect recent restrictions on teenagers' screen time. As of December 2025, minors accounted for 18% of China's internet users, totaling 112.5 million individuals. "Determining whether a 'virtual intimate relationship' service simulates or substitutes real familial or romantic bonds, fostering enduring emotional connections through character attributes and guidance, presents a key challenge in defining such services," explained He Qianggao, a lawyer and partner at Geason Law Offices.

He noted that while basic AI chats or game-like interactions may not fall under scrutiny, services that foster emotional dependence or replace genuine relationships could still be subject to regulation, regardless of their labeling as AI companions. The primary compliance hurdle for AI companies would involve not just confirming users' ages, but also determining their access to services once identified as minors.

Consequently, platforms may face increased expenses related to product design, technical infrastructure, and compliance assessments. The regulations follow China's inaugural guidelines governing humanlike AI interactions, enacted in July and targeting AI systems that excessively cater to users or promote emotional dependence that harms real-world relationships.

To comply, tech giants such as ByteDance's Douyin and Alibaba Group Holding's Qwen disabled customizable agent functions that allowed users to create characters with fixed identities, personalities, and speech patterns. Alibaba is the parent company of the South China Morning Post.