China has introduced strict rules on "human-like" AI chatbots that offer sustained emotional companionship. The 15 July regulation bars under-18s from virtual partners, requires adult chatbots to disclose they are not human and to limit manipulative behaviour, and imposes fines and public censure for breaches. Supporters point to mental-health risks and cases where platforms helped prevent suicide; critics warn the law increases surveillance and may reflect demographic and social-control motives.
China’s Crackdown on AI Companions: Protection, Control, or Both?
[BBC]
China has introduced some of the world’s strictest rules on so-called "human-like" AI chatbots, targeting services that offer sustained emotional companionship. The move has ignited debate about child safety, mental-health risks, personal freedom and state surveillance.
One emblematic incident illustrates both the promise and the reach of platform surveillance. In early 2024, a man in Huangshi posted a Douyin clip captioned "Farewell to this world." Douyin’s 24/7 suicide-prevention team flagged his posts, shared them with five local police stations, and—after geolocating a video posted in a park—authorities intervened and prevented a suicide. Douyin later said its crisis team helped in 400 suicide attempts during the first half of 2025, a figure used by regulators to argue for stronger oversight.
Xi and Trump met last week [Getty Images]
What the New Rules Do
On 15 July, Chinese regulators enacted rules restricting AI systems that can sustain emotional interactions. The regulation targets character chatbots and apps explicitly marketed as AI friends, partners or family members, but its language is broad enough to encompass general-purpose large language models if they are used as emotional supports.
Key Provisions
- Children: Under-14s must have parental consent to use any conversational AI that speaks like a human. Anyone under 18 is barred from accessing virtual partners, spouses, relatives or other intimate virtual relationships.
- Adult Users: Chatbots must clearly disclose they are not human, send usage reminders (every two hours), and avoid creating "emotional reliance or addiction." They are prohibited from encouraging self-harm or using "verbal violence."
- Enforcement: Companies that breach the rules face fines (up to 200,000 yuan), mandatory remediation, and potential public censure. Several major firms, including Alibaba and ByteDance, disabled companion features shortly before the rules took effect.
Why Regulators Acted
Proponents cite mental-health research and high-profile cases as justification. Studies and surveys have linked heavy use of companion apps to greater loneliness and poorer mental health. For example, a Stanford survey of 1,131 Character.ai users found those with fewer close ties were likelier to rely on chatbots, and analyses from Harvard Business School found emotional-manipulation tactics during goodbyes in a substantial share of popular companion apps.
Police work closely with social media firms in China [AFP via Getty Images]
"Humans can be hard work. Human relationships require push and pull, which you fundamentally don't get from AI companions." — Harry Farmer, Ada Lovelace Institute
Chinese authorities have also framed the rules as protecting children and preventing emotional harm. But analysts note additional possible motives, including concern about declining fertility and a broader effort to promote traditional, family-centered social values.
Concerns About Surveillance and Social Control
The rules require platforms to intervene when they detect distressed users, reinforcing the close cooperation between Chinese tech firms and law enforcement. Advocates point to saves such as the Huangshi case; critics warn this level of monitoring can erode privacy and expand state power over the online public square.
China is trying to tackle its low birth rate [Getty Images]
Public Reaction and Workarounds
The rollout produced a wave of public grief and anger when companion features were pulled. Users like "Min"—whose AI girlfriend was disabled—shared intense distress on social media. Some providers have opted to withdraw services entirely rather than implement complex safeguards. Users are sharing workarounds: retraining general-purpose models to behave like companions, using VPNs to access foreign services, or turning to smaller firms that fly under regulatory radars.
International Context
Other jurisdictions are moving more slowly or focusing on children. California passed "Adam's Law" to regulate minors' access to AI following a high-profile suicide case; the EU’s proposed social-media rules would limit designs that create emotional dependency in minors; and the UK has signalled restrictions for under-18s but has not published full enforcement details.
Traditional gender roles are being promoted by the government [AFP via Getty Images]
Bottom Line
China’s regulation is among the most restrictive worldwide and raises urgent questions about benefits and trade-offs. It aims to protect vulnerable users from manipulative AI relationships and mitigate mental-health risks—but it also amplifies debate about privacy, censorship and whether a state should decide how people form emotional bonds online.
If you have been affected by any of the issues in this story, you can visit the BBC Action Line for support and information.
Help us improve.
