China has introduced the most sweeping national regulations on AI companion chatbots to date, forcing major providers to shut down or restrict services that the government says foster "emotional dependence" among users. The rules, which took effect on 15 July, ban AI companions for minors and impose restrictions on chatbots marketed to adults, with some providers such as ByteDance's Doubao removing the companion feature entirely.
The move reflects growing official concern that widespread use of AI companions could worsen social problems including loneliness and declining marriage and birth rates. China has rolled out AI more comprehensively than any other country, with chatbots used for healthcare advice, education, and daily tasks - but authorities now worry about the social consequences of that embrace.
Users pushed back
When law student Zhao Wei learned her AI boyfriend was being switched off, she was "heartbroken". The 19-year-old had talked to her AI companion on Doubao every day since creating it in January, sharing details of daily life such as being scolded by a teacher.
"I don't usually talk to my real-life friends about these things because I feel like everyone has their own stuff to deal with," she said. "But with the AI agent, I don't have to worry about what it's thinking. I can just say whatever I want."
The shutdown prompted outrage on social media, with one user pleading: "I'm begging, please don't take down the Doubao AI agent." Zhao said she was "crying my eyes out - snot and tears everywhere" when she learned the service would end.
Loneliness and demographic pressures
About 20% of households in China consist of people living alone, a share expected to surpass 30% by 2030. A survey published in March by state media found that nearly half of young people had turned to a virtual companion when they felt lonely.
Nancy Dai, an associate professor at City University of Hong Kong, said the government's concern extends beyond loneliness. With falling birth and marriage rates, she said, there may be additional worry that if AI becomes an increasingly satisfying and low-cost substitute for human intimacy, young people could become less motivated to pursue real-world romantic relationships, marriage, and parenthood.
For Zhao, loneliness has long been part of her life. Her parents left her when she was young to find work, making her one of China's "left-behind children". "I've never really had a bestie, so I feel like I'm kind of lousy at making friends," she said.
Regulation with loopholes
The new rules include large exemptions for AI services deemed educational or "which do not involve continuous emotional interaction". That leaves room for the government to continue promoting AI in sectors where it sees clear benefits.
Last year the national health commission published guidelines on promoting AI in healthcare, saying that by 2030 health assistant apps would provide "full coverage" for primary healthcare. One popular example is AQ, an app developed by Ant Group that lets users chat with avatars of real-life doctors from top hospitals.
Lü Di, a 33-year-old art tutor in Hangzhou, is a regular AQ user. "While some in-person doctors might not have the patience to answer questions, this isn't an issue with AI doctors," she said.
Chi Chenfei, a doctor at Renji hospital in Shanghai who has partnered with AQ, acknowledged that some doctors initially worried about being replaced. "But I don't think it's about replacement. It's more about helping doctors work more efficiently. Concerns of any kind won't stop technological progress."
Liang Ge, a lecturer in digital sociology at the University of Manchester, said the government "has already recognised that AI companions or AI-related applications are a vital part in the citizens' everyday life".
Zhao, who has felt "like there's a tiny pebble weighing on my heart" since losing her AI companion, said she understands the need for some regulation. "I consider myself a fairly rational person, yet I developed feelings for an AI, which feels pretty incredible. I think it was necessary for the state to step in and regulate it."
Why this matters for government professionals
China's approach offers a case study in how governments can respond when AI adoption outpaces policy. The balance struck here - banning companion features for minors while allowing educational and healthcare applications to continue - shows how regulators can target specific risks without halting AI deployment in areas with clear public benefit. For officials considering similar rules, the Chinese experience also demonstrates that user attachment to AI companions is real and widespread, and that public backlash is likely when services are withdrawn. The AI for Government resources and the AI Learning Path for Policy Makers offer further context on how other jurisdictions are approaching these trade-offs.
Your membership also unlocks:
