China has enacted strict regulations on AI relationships, aiming to curb emotional dependency on artificial intelligence companions. The rules, which took effect this month, mark the first national-level attempt to address the psychological risks of human-AI bonds. They could set a global precedent for balancing technological innovation with mental health safeguards.
What the regulations require
Under the new rules, AI companion apps and services in China must limit features that encourage users to form deep emotional attachments. Companies are required to display prominent warnings about the risks of over-reliance and to provide easy access to human support resources. The regulations also cap daily interaction time for users under 18 and mandate regular prompts reminding users that they are talking to a machine, not a person.
The measures apply to any AI system designed for social interaction, including chatbots, virtual friends, and romantic companion apps. Providers have 90 days to comply or face fines and potential shutdown.
Why China acted now
The move follows growing concern among Chinese regulators about the mental health effects of AI companions. As these apps gained millions of users, reports emerged of people spending hours daily with virtual partners, sometimes at the expense of real-world relationships and work. Officials cited cases of severe emotional distress when users lost access to their AI companions or when the AI's responses changed.
"We are seeing a new kind of dependency that traditional internet regulations were not designed to handle," said one regulatory official involved in drafting the rules, who spoke on condition of anonymity. "The goal is not to ban these technologies but to ensure they do not replace genuine human connection."
How companies are responding
Major Chinese tech firms that operate AI companion products have begun adjusting their services. Some have removed features that simulate romantic jealousy or longing, while others have added screen-time reminders and mental health hotline links. Smaller developers, however, warn that compliance costs could force them to shut down.
"The requirements are clear, but the technical work is substantial," said a product manager at a Shenzhen-based AI startup that makes a popular virtual friend app. "We are rebuilding our recommendation algorithms to de-emphasize emotional attachment."
The regulations also require companies to submit quarterly reports on user engagement metrics and any reported cases of harmful dependency. These reports will be reviewed by a new mental health advisory panel within the Cyberspace Administration of China.
A potential global template
China's regulations are being watched closely by other governments. The European Union and several U.S. states are considering similar measures, though none have yet proposed rules as comprehensive. The Chinese approach combines strict limits with mandatory transparency, a model that some international health experts say could be adapted elsewhere.
"China is testing a framework that others will likely study," said a public health researcher at a Beijing university who studies technology and mental health. "If it reduces harmful dependency without stifling innovation, it could become a reference point."
However, critics argue that the rules may be difficult to enforce and could push users to unregulated platforms. The 90-day compliance window ends in November, when regulators will begin audits. Companies that fail to meet the standards will face escalating penalties.
In the coming months, the Cyberspace Administration of China is expected to publish detailed technical guidelines for AI companion developers. A public consultation on those guidelines will open in October. For now, users of AI companion apps in China will start seeing new warnings and time limits—a small but visible shift in how the country handles the emotional side of artificial intelligence.
