China’s Companion-AI Law: Emotional Chatbots Get Their Own Bucket
China’s anthropomorphic AI interaction rules took effect July 15—targeting continuous emotional companions with youth and disclosure duties, while exempting ordinary customer service, Q&A, and work tools. Product taxonomy, not a blanket generative-AI ban.
If your UX optimizes for relationship and affect, you are in a different legal bucket than if you optimize for tasks. China’s July 15 effective date makes that taxonomy hard law for companion products—while productivity and research agents stay under older generative-AI and content rules.
China’s Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interaction Services came into force July 15, 2026. Issued April 10, 2026 by multiple central authorities (including the Cyberspace Administration of China, NDRC, MIIT, Ministry of Public Security, and SAMR), the rules target continuous emotional interaction services that simulate a natural person’s personality, thinking, and communication style. Ordinary customer service, knowledge Q&A, work assistants, education, and research tools without continuous emotional interaction are out of scope.
What the regime covers
From the published measures and English-language legal/tech analyses (China Law Translate, Geopolitechs, compliance briefs):
| Theme | Direction |
|---|---|
| In scope | Public-facing services in China that deliver sustained emotional / anthropomorphic interaction via AI |
| Out of scope | Intelligent customer service, knowledge Q&A, work assistants, education/learning, scientific research without continuous emotional interaction |
| Youth | Heightened duties: age recognition, guardian consent patterns, usage limits, bans on designs that replace social interaction for minors (core draft/final themes) |
| Content & security | Prohibitions aligned with broader Chinese content law; provider responsibility for interactive content |
| User disclosure | Human-like agents expected to avoid deceptive “I am a real person” framing; psychological-dependence risks addressed in design duties |
| Effective date | July 15, 2026 (Article 32 of the measures) |
The package is one of the world’s most product-specific AI rules: not “all generative AI,” but the companion surface regulators associate with mental health, addiction, and minor safety—after high-visibility companion apps in the Chinese market.
Limits
- English summaries and legal translations are secondary to the PRC text; cite measures for compliance work.
- Exact enforcement practice (fines, takedowns) will show in post-effective cases, not the day-one statute alone.
- Out-of-scope productivity tools can still fall under older generative-AI and content regimes.
Sources
- PRC multi-agency Interim Measures for the Management/Administration of Anthropomorphic AI Interaction Services (issued April 10, 2026; effective July 15, 2026).
- Geopolitechs / China Law Translate English analyses of the measures and draft history.
- Compliance summaries noting minor protections and emotional-interaction scope (e.g., Licentium, ITTC Network briefs on the July 15 effective date).