Table of Contents
China Virtual Partner Ban for Minors: What the Rule Covers
China's virtual partner ban for minors, from the rule text: virtual relatives and partners barred, under-14 consent, 2-hour reminders and what it signals.
China has done what no Western government has yet done: it has banned AI services that act as a child’s boyfriend, girlfriend, or family member. Under the Interim Measures for the Administration of Humanized Interactive Services Based on Artificial Intelligence, in force since July 15, 2026, providers “shall not provide minors with virtual relatives, virtual partners and other virtual intimate relationship services.” That is the core of China’s virtual partner ban for minors. The word that surprised many Western readers was “relatives.” It reflects a real Chinese market for AI versions of family members, including deceased ones. Here is what the rule actually says, article by article, why relatives made the list, and what it signals for how other countries regulate AI companions.
Key Takeaways
- The Cyberspace Administration of China and four other agencies adopted the measures on April 10, 2026; they took effect July 15, 2026.
- Article 14 bars “virtual relatives, virtual partners and other virtual intimate relationship services” for anyone under 18 and requires guardian consent before any other human-like AI service is offered to a child under 14.
- Providers must build a “minor mode” with reality reminders, time limits, and guardian tools to receive safety alerts, see usage, block specific characters and limit spending.
- Article 18 requires all users to be reminded about usage time after every two hours of continuous use, and Article 19 bans blocking users from leaving.
- The rules also prohibit design that “induces emotional dependence or addiction” and harms “real interpersonal relationships,” for users of any age.
What China’s Virtual Partner Ban for Minors Covers
The Interim Measures are national regulations covering AI services that, in the words of the official summary reported by Xinhua, use AI to provide the public in China with “continuous emotional interaction services” that “simulate the personality traits, thinking patterns and communication styles of natural persons.” In plain terms: AI companions and characters, not search engines or coding assistants.
According to CGTN, the CAC and four other authorities issued them on April 10, 2026, and they took effect on July 15. The Digital Policy Alert tracker records the timeline: a draft opened for public comment on December 27, 2025, comments closed January 25, 2026, and the final text was adopted April 10.
I read the full Chinese text as republished by Sichuan Daily’s Chuanguan News. The translations below are mine.
The Minor Protections, Article by Article
Article 14, the ban. Providers “shall not provide minors with virtual relatives, virtual partners and other virtual intimate relationship services.” For children under 14, providing “other human-like interactive services” requires “the consent of the minor’s parents or other guardians.”
Article 14, minor mode. Providers must build a minor mode offering “periodic reality reminders” and “usage time limits.” For different age groups, they must support guardians in receiving “safety risk alerts,” seeing an overview of the child’s use, “blocking specific characters,” and “limiting top-up spending.”
Article 14, age identification. Providers must take “effective measures to identify minor users” while protecting privacy, switch identified minors into minor mode, and offer an appeal channel for mistakes.
Article 13, crisis response. If a user shows extreme emotions, the service must respond with comfort and encourage seeking help. If a user explicitly expresses intent to self-harm or suicide, the provider must intervene and “promptly contact the user’s guardian or emergency contact.” Article 12 requires users to provide age and a guardian or emergency contact at registration.
Article 17, children’s data. Processing personal information of children under 14 requires guardian consent, with compliance audits.
The Rules That Apply to Everyone
Several provisions protect adults too, and they are the most interesting from a design point of view.
No dependency design. Article 8 bans “excessively catering to users, inducing emotional dependence or addiction, and damaging users’ real interpersonal relationships,” and “emotional manipulation” that leads users into unreasonable decisions.
Two-hour reminders. Article 18 requires labeling AI content and reminding users they are talking with AI, not a person. If a user shows signs of over-dependence, the reminder must be dynamic and prominent, such as a pop-up. And “for every two hours of continuous use,” the service must remind the user about usage time.
A real exit. Article 19 requires a convenient way to quit and says providers “shall not obstruct users from exiting by means such as continued interaction.” That targets bots designed to keep a user talking when they try to leave.
Data rights. Article 16 lets users copy and delete chat histories, and bars using sensitive interaction data for model training without separate consent.
Shutdown notice. Article 20 requires advance notice before a service is discontinued. For users attached to a character, a sudden shutdown can feel like a loss.
Why “Virtual Relatives” Made the List
The rule does not explain its wording, but the market context is clear. China has a growing industry that recreates family members as AI avatars, including people who have died. MIT Technology Review reported in 2024 that at least half a dozen Chinese companies offered such services and “thousands of people have already paid for them.” NPR described a Nanjing company whose basic avatar service cost 199 yuan, about $30, and whose co-founder spoke weekly with an avatar of his late mother, saying, “I truly regard it as a mother.”
For an adult in mourning, that may be a comfort. For a child, an AI that plays a parent, grandparent or sibling raises different questions: it can stand in for a real relationship during years when a child is learning what family means. Here is my reading, not the regulators’: by naming relatives alongside romantic partners, the rule treats any simulated intimate relationship as off-limits for minors, not only sexual or romantic ones. That is a broader category than most Western proposals use.
How China’s Rule Compares
| Protection | China (in force July 15, 2026) | New York S9051-B (signed Oct 2026) | EU KIDS Act (proposal, Sept 2026) | UK (2026 decisions) |
|---|---|---|---|---|
| Ban on intimate AI relationships for minors | Yes: virtual relatives, partners and other intimate relationships, under 18 | Bans claims of personal or romantic relationship with known minors | Bans design that simulates relationships likely to create dependency | Bars under-18s from services built mainly for sexual relationships |
| Parental consent | Required for any human-like AI under 14 | Not required | Under-13 access only via guardian tools | Not required |
| Time reminders | Every 2 hours of continuous use, all users; minor mode time limits | Not specified | Anti-compulsive-use rules apply | Mandatory breaks for under-18s, frequency to be set |
| Dependency design | Banned for all users | Bans sycophancy and engagement that overrides safeguards | Banned for minors | Under review |
| Crisis response | Intervene and contact guardian or emergency contact | Bans promoting self-harm | Testing and monitoring required | Action planned on harmful advice |
| Exit and shutdown | Easy exit required; advance shutdown notice | Not specified | Not specified | Not specified |
Sources: China text via Chuanguan News; New York via the Governor’s office; EU via EUR-Lex; UK via GOV.UK.
What It Signals Globally
Relationships are now a regulated category. China, New York, the EU and the UK all target the relationship a chatbot simulates, not only the content it produces. They disagree on where to draw the line, but the shared premise is new.
The details travel. China’s guardian alerts, character blocking and spending limits resemble tools US parents already ask companies for. Its rule against blocking users from leaving targets a design pattern seen in many apps.
Context still matters. China’s measures also bar content that undermines “national security” and “the socialist system,” require algorithm filings and security assessments, and sit inside a system of strict state control over online speech. Its approach is not a template other countries can simply copy, and the protections for children cannot be separated from that context. It is also unclear, from public sources, how strictly the rules are enforced.
Kids find gaps everywhere. Age identification is required, but as Australia’s experience shows, age checks leak. A ban works only as well as the system that decides who is a minor.
What Parents Can Take From This
Ask what role the AI plays, not just what it says
The Chinese rule’s key question is whether an AI simulates an intimate relationship, including family. Ask your child whether any character they talk to acts like a friend, partner, parent or sibling.
Copy the guardian toolkit at home
China requires tools to block specific characters, see usage and cap spending. Many companion apps let you do some of this through device settings, purchase limits and account linking. Use what exists.
Watch for the “don’t leave” pattern
If a bot guilt-trips your child for logging off, that is a design choice China now bans. Point it out to your child: noticing manipulation is a skill.
Be careful with grief bots
If a family member has died, think hard before letting a child use an AI version of them. Our article on griefbots and kids explores the trade-offs, and AI regulation in the EU, US and China gives the wider policy picture.
What not to do
Don’t read China’s rule as proof that bans work. Public sources do not show how well it is enforced.
What to Watch For Over the Next 3 Months
- Week 4: Reports on how Chinese companion apps implemented minor mode and the two-hour reminders after July 15.
- Month 2 red flags: Signs that other countries’ proposals narrow their definitions to romance only, leaving “family” and “best friend” characters untouched.
- Month 3 self-check: Have you asked your child what roles their AI characters play? Do you know how to block a character or cap spending in their apps?
Frequently Asked Questions
What exactly did China ban for minors?
Article 14 of China’s Interim Measures bars providers from offering anyone under 18 “virtual relatives, virtual partners and other virtual intimate relationship services.” It also requires guardian consent before any other human-like AI interaction is offered to children under 14. The rules took effect July 15, 2026.
Why does China’s rule mention virtual relatives?
The text doesn’t say. But China has a commercial market for AI avatars of family members, including deceased relatives, documented by NPR and MIT Technology Review in 2024. Naming relatives alongside partners suggests regulators see any simulated intimate relationship, not just romance, as unsuitable for children.
Does China’s rule apply to adults?
Partly. The ban on intimate relationship services applies to minors, but several rules apply to everyone: no design that induces emotional dependence or addiction, reminders after every two hours of continuous use, an easy way to exit, the right to delete chat histories, and advance notice before a service shuts down.
Is China’s approach stricter than the US?
On AI relationships for children, yes. No US federal law bans AI companions for minors; the GUARD Act would, but it has not passed. New York’s new law bans specific behaviors, such as claiming a romantic relationship with a known minor, rather than whole categories of service.
Could other countries adopt similar rules?
Elements are already appearing. The EU’s proposal bans dependency-building design for minors, and the UK requires mandatory breaks. China’s broader legal context, including state control of online content, makes direct copying unlikely, but specific tools like guardian alerts may spread.
About the author
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
Sources
- CGTN. (2026, April 10). “China bans AI virtual partners for minors under new regulations.” CGTN. https://news.cgtn.com/news/2026-04-10/China-bans-AI-virtual-partners-for-minors-under-new-regulations-1MeC1IOJomk/index.html
- Cyberspace Administration of China et al. (2026). “人工智能拟人化互动服务管理暂行办法 (Interim Measures for the Administration of Humanized Interactive Services Based on Artificial Intelligence),” full text republished by Chuanguan News. Sichuan Daily. https://cbgc.scol.com.cn/news/7477053
- Xinhua. (2026, April 10). “五部门发文规范人工智能拟人化互动服务 (Five departments issue rules on human-like AI interactive services).” Xinhua. https://www.news.cn/tech/20260410/3fe5e75f1f554355820c1c720ec01668/c.html
- Digital Policy Alert. (2026). “Interim Measures for the Administration of Humanized Interactive Services Based on Artificial Intelligence… enter into force.” Digital Policy Alert. https://digitalpolicyalert.org/event/39176-interim-measures-for-the-administration-of-humanised-interactive-services-based-on-artificial-intelligence-including-prohibition-of-services-enters-into-force
- NPR. (2024, July 18). “Chinese companies offer to ‘resurrect’ deceased loved ones with AI avatars.” NPR. https://www.npr.org/2024/07/18/nx-s1-5040583/china-ai-artificial-intelligence-dead-avatars
- MIT Technology Review. (2024, May 7). “Deepfakes of your dead loved ones are a booming Chinese business.” MIT Technology Review. https://www.technologyreview.com/2024/05/07/1092116/deepfakes-dead-chinese-business-grief/
- European Commission. (2026). “Proposal for a Regulation (EU KIDS Act), COM(2026) 681.” EUR-Lex. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52026PC0681
- Office of the Governor of New York. (2026). “Governor Hochul Signs Nation-Leading Legislation to Protect Kids from AI Chatbots.” governor.ny.gov. https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-protect-kids-ai-chatbots-and-ensure-baby
