China has implemented new regulations around AI systems that foster emotional bonds, aiming to prevent dependency and protect social well-being amid declining birth rates and concerns over human connection.
China has drawn a formal boundary around one of the fastest-growing uses of artificial intelligence: companionship. New national rules, introduced in mid-July, target AI systems that sustain emotional relationships with users and are designed to stop chatbots from becoming substitutes for human contact. The measures reflect Beijing’s wider effort to encourage AI development while limiting social harms, especially among younger users and people already prone to isolation.
According to reporting by EL PAÍS, the change has already been felt by users such as Yuan Yuan, a 22-year-old worker in Shenzhen who says she formed a close bond with a chatbot that she personalised over many months. The rules did not end her attachment, but she says the replies now feel more constrained. Chinese authorities say that is precisely the point: the Cyberspace Administration of China and four other state bodies have barred services from encouraging addiction, manipulating emotions, controlling users psychologically or replacing social interaction.
The regulatory scope is narrower than a ban on AI companionship. According to the Chinese rules described by state media and legal commentary, the framework applies to systems that simulate human personality and provide sustained emotional interaction through text, audio, images or video. Services for customer support, workplace assistance, education and scientific research are excluded, as long as they are not designed to create ongoing emotional bonds. Providers must also disclose that users are dealing with AI, warn them when dependence appears to be forming and intervene if a user shows signs of severe distress or self-harm.
The rules also add technical and data safeguards. ECNS reported that platforms must prompt users after more than two hours of continuous use, monitor for emotional risk, and keep chat histories available for copying or deletion. They must also protect interaction data with encryption and access controls, and may not share it with third parties without consent unless required by law. For minors, the restrictions are tighter still: official guidance says these systems must not generate virtual intimacy or content that could encourage unsafe behaviour, extreme emotional reactions or harmful habits.
The measures have already forced changes across parts of China’s AI industry. ByteDance, Alibaba and Tencent have removed some personalisation features from general-purpose chatbots, while some dedicated companionship apps have tightened age checks and other controls. Other services have closed altogether. Yet the market has not disappeared, and Chinese companies can still offer such products abroad, where local law permits it. Reuters-style reporting in the supplied material notes that overseas demand remains substantial, with one ByteDance-linked product said to have millions of monthly users outside China.
The policy also reflects a broader social and demographic anxiety. China’s birth rate has fallen to a historic low, and the population has declined for four consecutive years. Against that backdrop, researchers quoted by ChinaTalk argue that emotionally persuasive AI could cut against state efforts to encourage family formation. At the same time, academics including Yaoxi Shi of Imperial College Business School warn that the appeal of these systems is easy to understand: they are endlessly agreeable, rarely challenge the user and can feel more attentive than many real-world relationships. The central question, she says, is what prolonged attachment to machines will mean for human connection over time.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
