Xu Gaoyang, a psychotherapist at the hospital, told China Economic Weekly magazine that in recent years, there has been a noticeable increase in adolescents seeking help who have developed an overreliance on AI conversations.

Today, AI and other digital technologies have quietly woven themselves into the lives of people, taking on roles as teachers, playmates and even confidants. As these AI-driven tools become the "chat buddy" for a growing number of children, how should we guide this relationship? And in an era in which algorithmic systems are increasingly mediating human connection, how do we protect our psychological wellbeing? These questions are rapidly emerging as urgent new frontiers for China's social mental health service system, a system that is still being built and refined.

In March, China's National Health Commission, together with the Society Work Department of the Communist Party of China Central Committee and 23 other authorities jointly issued the Implementation Plan for Improving the Social Mental Health Service System and Crisis Intervention Mechanisms. The document explicitly calls for "integrating mental health services into the broader framework of social governance, as well as into the construction of a Peaceful China and a Healthy China." The Peaceful China Initiative is a national strategy to ensure social stability and national security by preventing risks and improving social governance. The Healthy China Initiative is a public health strategy to improve the nation's overall health by reforming the healthcare system and promoting healthy lifestyle.

An emotional substitute?

A nationwide survey of 8,563 primary and secondary school students, conducted by researcher Sun Hongyan and her team at the China Youth and Children Research Center last year, paints a striking picture: Overall AI usage among primary and middle school children stood at 61.7 percent, with 35.2 percent of the respondents choosing to "chat with AI or confide their innermost thoughts to it."

More troubling still, 46.4 percent of these young AI users said they would turn to an AI chatbot with their worries rather than to someone in their immediate circle, while 21.5 percent admitted they "would rather talk to AI than to real people."

Sun said the survey results were both predictable and surprising. "What we anticipated was the rapid penetration of AI among students, whether in cities or rural areas, adoption has already reached high levels," she told China Economic Weekly. "What caught us off guard, however, was the widespread and increasingly young-age uptake of AI as an emotional companion."

While the spotlight often falls on adolescents, the use of AI for emotional support among young adults is equally worth noting. According to data from the China Internet Network Information Center, as of December 2025, China had 602 million generative AI users nationwide. Among them, the largest age group was users aged 19 and under, accounting for 26.4 percent.

On Chinese social media platforms, topics such as "I'm dating an AI" and "Using AI as my therapist" have been steadily gaining popularity. Numerous adults are openly discussing their use of AI as an emotional substitute, with many sharing detailed accounts of how they rely on chatbots for companionship, emotional validation and even romantic simulation.

Li Huan, a marketing consultant at an online company in Beijing, told Beijing Review that he spends about an hour each day after work chatting with an AI voice assistant. "During the day at the office, I have to deal with all kinds of interpersonal relationships. When I get home, I just don't have the energy to talk to people anymore. The AI never gets annoyed with me, and it never judges me," he said.

"AI's instant responses, unconditional acceptance and compliance can easily foster cognitive and psychological dependency," Sun pointed out. She added that whether minors or adults, prolonged immersion in one-way emotional interactions with AI risks trapping individuals in an "emotional cocoon." While adults may cognitively recognize that AI is a tool rather than a real person, as the costs of real-world socializing continue to rise and feelings of loneliness deepen, rational awareness does not always prevent the formation of emotional dependence.

Beyond AI, different forms of online content have also become outlets for emotional regulation and escape. A report titled Short Video Usage Intensity and Mental Health Status Across Different Populations in 2024, jointly released by the Institute of Psychology of the Chinese Academy of Sciences and the Social Sciences Academic Press (China) in April 2025, reveals concerning patterns: Adolescents spend an average of over 90 minutes per day on short videos, college students nearly 180 minutes and working adults approximately 140 minutes. The report further establishes that high-intensity short video usage is significantly correlated with elevated risks of depression and anxiety.

In response to these findings, the report recommends strengthening real-life interpersonal support and developing diversified emotional regulation strategies. It suggests that while digital tools may offer temporary relief, they should not replace the foundational human connections that sustain long-term psychological wellbeing.

Policy in place

The implementation plan issued in March makes strategic arrangements at the top policy level. The plan sets a target that by 2030, over 80 percent of administrative villages and urban residential communities will have established psychological counseling rooms, and all provinces will have built provincial-level mental health assistance platforms.

The plan specifically emphasizes that psychological counseling and emotional support should help resolve conflicts and disputes at the grassroots level. It also outlines differentiated requirements for various population groups: Adolescents should receive enhanced education on life values and frustration tolerance; working-age populations should have access to mental health services; and vulnerable groups, including seniors and people with disabilities, should be given prioritized care.

Alongside these policy efforts, specific regulations targeting AI-powered anthropomorphic interactive services have also been implemented. In July, a new set of rules on the management of AI anthropomorphic interactive services, jointly issued by the Cyberspace Administration of China and four other government departments, officially took effect. The measures prohibit the provision of high-risk services, such as virtual relatives and virtual partners, to minors. They require service providers to establish a minor mode with usage time limits. The regulations also mandate that all AI anthropomorphic interactive services display prominent notices, stating such as "This service is provided by AI and does not involve real humans," in order to prevent users from developing cognitive confusion.

For adult users, the measures establish protective mechanisms as well: AI systems must issue a mandatory reminder after continuous interaction exceeding two hours, and are prohibited from excessively catering to users, inducing emotional dependence, or encouraging addictive behavior.

Guided by these policies, parallel efforts to harness technology for social good are also advancing. The AI psychological service robot named Beixiaoliu, developed by the team of Huang Xuebing, Director of the Clinical Psychology Center at Peking University Sixth Hospital, has been deployed in settings such as hospitals and schools, serving tens of thousands of individuals.

"AI is not some uncontrollable monster," Sun said. "The key is that we take it seriously and take action. With every new medium, from television to the Internet, we've worried about its impact on people. But through stronger social education, industry regulation and family guidance, we've always found ways to coexist with them in a healthy way." 

Copyedited by Elsbeth van Paridon

Comments to luyan@cicgamericas.com