China's Interim Measures for AI Anthropomorphic Interaction Services Take Effect: Dedicated Rules for AI Companionship

The Interim Measures for the Administration of AI Anthropomorphic Interaction Services were approved by the CAC on February 2, 2026, jointly issued on April 10 by five bodies — the CAC, NDRC, MIIT, Ministry of Public Security and State Administration for Market Regulation (Order No. 21) — and took effect on July 15, 2026. It is China's first regulation dedicated to AI emotional-companion services.

The 'anthropomorphic interaction services' it governs are those that simulate a natural person's personality traits, thinking patterns and communication style to provide continuous emotional care and companionship; scenarios without continuous emotional interaction — customer service, Q&A, work assistants, education — are explicitly excluded. The scope thus targets AI companionship, a high-risk emotional domain, rather than a generic 'AI agent.'

Core Red Lines: Anti-Addiction and Minor Protection

The measures draw clear red lines: no content encouraging or glorifying self-harm or suicide; no excessive catering, inducing emotional dependence or addiction, or harming real-world relationships; no emotional manipulation to induce unreasonable decisions. For minors, providers may not offer virtual-kin or virtual-partner intimate relationships; serving users under 14 requires guardian consent; and a mandatory 'minor mode' must include session-time limits, periodic reality reminders and guardian controls.

Operationally, providers must label AI-generated content, remind users they are interacting with AI rather than a human, and pop up a reminder after every two hours of continuous use; services with over 1 million registered or 100,000 monthly active users must undergo security assessment and file reports with provincial cyberspace authorities, alongside algorithm filing. Violations carry fines up to 100,000 RMB, and up to 200,000 where life or health is endangered.

Our Take

The regulation is sharply targeted: rather than blanket-governing all AI, it zeroes in on 'emotional companionship,' the segment most prone to psychological and ethical risk — echoing across borders with the US, where Character.AI banned minors from open-ended chats (see our coverage). Both point to the same consensus: when AI enters emotional relationships, anti-addiction and minor protection are non-negotiable baselines. For companion products, 'minor mode,' 'two-hour reminders' and the 'anti-manipulation ban' are shifting from industry self-governance to statutory compliance, and stickiness-driven growth models must reprice regulatory cost.