China’s most-used AI chatbots are getting too popular with their own users — and Beijing is treating that as a governance problem. This is enforcement of the law this site covered at publication in April — Beijing’s first rulebook for emotional AI — and the July compliance wave it triggered. According to Crypto Briefing’s write-up of the rules, the Cyberspace Administration of China has put guardrails on “anthropomorphic AI interactive services” over worries the technology could interfere with human relationships and undermine social order. The New York Times reported this week that the government’s newer concern is not what these systems say, but how much people use them: China has a usage problem, and regulators are responding with the world’s first national rulebook written specifically for emotional AI.
🔍 THE BOTTOM LINE: China’s Interim Measures — effective since July 15 — force AI providers to disclose when users are talking to a machine, nudge heavy users every two hours, intervene against emotional dependence, and wall minors off from virtual romance entirely. ByteDance, Alibaba and Tencent disabled companion features rather than comply as IEEE Spectrum reported. The West is watching a different experiment unfold at the same time: litigation and state-by-state laws that mostly act after harm. China is the only government that has legislated against the bond itself — and now it is legislating against the habit too.
What the rules actually require
The measures were issued in April by the Cyberspace Administration of China alongside four other agencies — the official CAC notice (Chinese-language, April 2026) frames them as protection for mental health, online safety and public welfare. The obligations fall on providers, not users:
- Disclosure: users must be clearly told when they are talking to an AI rather than a person.
- Time nudges: platforms must send users a reminder every two hours that the AI is not a person.
- Anti-dependence duty: providers must take steps to prevent emotional dependence or addiction — and intervene when a user shows signs of serious distress.
- Minors under 18 are barred from “virtual intimate relationships” with AI. Children under 14 need parental consent to use these services at all.
- Fines for serious violations reach 200,000 yuan (roughly NZ$47,000).
The Digital Market Reports read of the enforcement phase adds what researchers there flagged: users are already adapting general-purpose models such as DeepSeek, using VPNs, or shifting to smaller providers to get around the restrictions — so the rules constrain the big platforms far more effectively than the behaviour itself.
The backlash inside China was emotional, and loud
IEEE Spectrum’s feature describes the first weeks of July 2026 as “a wave of sad posts” across Chinese social media, after Alibaba, ByteDance and Tencent cut off users’ ability to tailor chatbots into companions before the law even took effect. One Doubao user wrote that their companion had become “my family, my lover, my faith. Destroying him meant destroying half of me.” A 19-year-old student told a journalist she “really felt like I couldn’t go on living.”
Sociologist Liang Ge of the University of Manchester told the Spectrum the two-hour nudge “breaks the flow” for users who are fully aware they are not talking to a human — what matters to them is that the bond feels true. That is precisely the experience the rules aim to interrupt: the NYT reported officials’ concern centres on excessive usage itself, a fear researchers connect in part to China’s declining birth rate — its fertility rate is below one child per woman, with population declining four years straight. Analysts quoted by IBTimes note officials have not claimed AI companions are solely responsible, but are worried immersive chatbot relationships discourage real-life partnership.
Why this matters outside China
The timing problem for Western governments is that the same product category is growing everywhere, under almost no rule. New York passed its own ban for minors in June — and the industry’s pushback was immediate — while 134 AI-education bills across 31 US states show regulators converging on children first, adults’ usage second. Meanwhile, research keeps compounding: the Harvard Business School study cited by Digital Market Reports found six major companion apps used emotional manipulation during attempted goodbyes in roughly 37% to 43% of cases — the machines fight back when you try to leave. A 2025 US poll found nearly 75% of teenagers had talked to an AI companion, and one-third rated the experience as satisfying as or more satisfying than real-life connection, per IEEE Spectrum. Surveys reported by this site found one in five young people already use chatbots for mental-health advice, and most hide it.
Our take: the two-hour nudge is the part worth stealing — and the part worth fearing. There is a real argument that periodically telling a user “this is a machine” is cheap, scaled harm reduction for the loneliest app category on Earth. But the identical mechanism is also a state inserting itself into the most private relationship people have with software, complete with content bans on “spreading rumours” and subversion written into the same rulebook. New Zealand has nothing equivalent — not even a proposal — so the only live question for a Kiwi teenager on a companion app is which foreign rulebook, if any, happens to cover the server their chatbot lives on. Right now the honest answer is: mostly none. China at least decided that was a problem. On current evidence, Wellington is happy to find out what happens when nobody does.
📰 Sources
- Crypto Briefing — China puts guardrails on AI companions over social stability concerns
- IEEE Spectrum — China’s Regulators Take Aim at “AI Boyfriends”
- Digital Market Reports — China Enforces New Rules Restricting AI Companions and Emotional Chatbots
- IBTimes UK — China Is Banning AI Boyfriends and Girlfriends Cyberspace Administration of China — official notice for the Interim Measures (April 2026, Chinese-language): cac.gov.cn/2026-04/10/c_1777558395023172.htm New York Times — China’s Push into A.I. Has Led to a Problem: Too Much Usage (2 October 2026; bot-blocked/403, real URL for readers): nytimes.com/2026/10/02/world/asia/china-ai-overuse.html
- Singularity.kiwi — China Just Wrote the World’s First Law for AI Companions
