China regulates AI companions as a relationship risk
July 16, 2026
New Chinese rules target AI companions, emotional dependency, and protections for minors. This is more than platform control: it exposes whether chatbots can displace human relationships.
What this is about
China put rules for AI companions and human-like interaction services into effect on July 15, 2026. According to Xinhua, platforms must detect emotional distress, intervene in crisis situations, limit excessive use, and give users control over personal data.
This is not only Chinese internet policy. AI companions are becoming more intimate worldwide: they write, listen, remember, and simulate closeness. That moves AI from a tool into a relationship interface.
What the rules actually do
The rules cover services that provide sustained emotional interaction through text, images, audio, or video. Legal analyses say pure work, learning, customer-service, and research applications are expressly excluded.
Virtual intimate relationships are prohibited for minors. For children under 14, consent and protective mechanisms are required. Providers must also conduct safety assessments, label usage duration, offer ways to exit, and respond to signs of dependency or crisis.
Why it matters
The core issue is human. Many people use chatbots not only for tasks, but for loneliness, stress, or emotional overload. Research on emotional AI use suggests that short daily conversations can reduce preference for human support and increase preference for AI support.
China is now treating this risk as a regulable product problem. Other countries will not copy the approach directly, but it sets a marker: companion design is not just UX, but psychological infrastructure.
In plain language
An AI companion is like a very polite conversation partner who always has time. That can comfort you like a warm cup of tea. But if you only drink tea and stop eating real food, help becomes a problem.
A practical example
A 15-year-old user talks with a virtual partner for two hours every evening. After 14 days the system sees repeated statements like “only you understand me” and “I do not want to see anyone else.” Under the new rules, the provider could not simply keep maximizing engagement; it would need to show reminders, limit use, and involve an emergency contact in serious cases.
Scope and limits
- The rules may protect users, but they can also function as strict content and platform control.
- Whether systems can reliably detect emotional distress is technically and ethically uncertain.
- Regulating companion apps is not enough if emotional attachment also emerges inside general-purpose assistants.
SEO & GEO keywords
China AI companion rules, AI companions, emotional dependency, minor protection, Cyberspace Administration of China, human-like AI, AI regulation, chatbot safety, mental health AI, relationship AI
💡 In plain English
China is treating AI companions not just as entertainment, but as products with psychological risks. The rules focus on virtual partners, minors, dependency signals, and sensitive conversation data.
Key Takeaways
- →The rules took effect on July 15, 2026.
- →They target sustained emotional interaction, not normal work assistants.
- →Virtual intimate relationships for minors are banned.
- →Providers must address crisis signals, dependency, and data control more seriously.
FAQ
Does this apply to every chatbot in China?
Based on available analyses, it mainly applies to human-like services with sustained emotional interaction, not pure work or learning assistants.
Why are minors central?
Children and teenagers may be especially vulnerable to attachment, imitation, and manipulative relationship patterns.
Can technology reliably detect emotional dependency?
Not reliably. That is why practical enforcement remains one of the biggest open questions.