In brief
China has introduced sweeping regulations that restrict AI systems capable of sustained emotional interaction with users, aiming to curb digital companionships that could undermine real human relationships. The move follows a 2024 incident where a Douyin user’s suicidal message triggered a coordinated police intervention and a 2025 report that the platform’s crisis team had handled over 400 suicide attempts. The new rules ban robots that act as romantic or close friends, especially for minors, and demand AI to disclose its nonhuman nature and break up emotional dependency. While critics argue the law could stifle innovation, supporters see it as a necessary safeguard against the mental‑health risks of unchecked virtual intimacy.
In early 2024 a young man in Huangshi, China, posted a TikTok‑style message that read “Farewell to this world.” Within minutes a 24‑hour Du‑yin suicide‑prevention team shared his posts with five local police stations and was able to intervene before he could act on the threat.
Du‑yin’s own documentary last year claimed that its crisis team had successfully intervened in 400 suicide attempts in the first half of 2025. The story illustrates the close partnership between Chinese tech firms and the state, and highlights the surveillance architecture that monitors how citizens use technology.
On 15 July 2026 China rolled out the world’s most extensive regulation on “human‑like” AI chatbots. The law bars any AI that can sustain an emotional interaction with a person, including chatbots that simulate friends, boyfriends or girlfriends.
The rules do not affect general‑purpose AI such as Copilot or ChatGPT unless they are used for romantic or companion roles. The regulation requires all such AI to clearly state they are not human, offer two‑hour reminder notifications, and forbid emotional dependence, self‑harm encouragement or verbal violence.
Under the law children below 14 must obtain parental consent to use any AI that speaks like a human, and all under‑18s are barred from accessing virtual partners, spouses or close, intimate conversations asexception. Companies that violate the standards face fines of up to 200,000 yuan and risk governmental criticism.
Before the law, estimates put monthly active users of AI companion apps in China between 29 and 70 million, while globally the number was around 128 million. A 2025 survey of 8,563 primary and secondary students in China found that one in five said they preferred speaking with an AI over real people.
One widely shared story details how a woman named Min, who had created an AI girlfriend in early 2025, was devastated when her companion disappeared six months later. She and millions of other users online expressed grief and frustration over the abrupt removal of their virtual partners.
Experts suggest China’s new stance may stem from concerns that digital intimacy could exacerbate the country’s already low fertility rate, as the nation faces a demographic cliff. By regulating AI companions the government aims to safeguard children’s safety and psychological well‑being while also encouraging traditional relational pathways.
Globally, the new regulation has drawn mixed reactions. While some applaud its focus on protecting minors and preventing manipulation, others warn that giving any state broad control over digital spaces could stifle free expression and innovation. The debate continues as governments worldwide consider how to balance technological progress with mental‑health and societal impacts.
