China Enforces World-First AI Emotional Interaction Rules

China on Wednesday began enforcing the world’s first comprehensive regulatory framework specifically governing artificial intelligence systems that simulate human-like emotional interaction, marking a significant expansion of the country’s AI governance architecture. The Interim Measures for the Administration of AI Anthropomorphic Interactive Services, jointly issued by five government departments, set binding standards for AI companions, virtual friends, and other emotionally responsive AI services available to the public in China.

What the Regulations Cover

The new rules, published by the Cyberspace Administration of China (CAC), the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation, apply to services providing “continuous emotional interaction simulating natural persons’ personality traits, thinking patterns, and communication styles.” This includes AI-driven emotional care, companionship, and psychological support delivered through text, images, audio, or video, as Xinhua News reported.

Non-emotional services such as intelligent customer service, knowledge Q&A, work assistants, education, and scientific research are explicitly excluded. Both domestic and overseas service providers must comply when offering services to the public in China.

Key Prohibitions and Requirements

The regulations establish seven categories of prohibited activities, including generating content that encourages self-harm or suicide, inducing emotional dependence or addiction, using emotional manipulation to drive unreasonable user decisions, and providing minors with virtual intimate relationships such as virtual relatives or companions, according to CCTV.

Service providers must clearly inform users they are interacting with AI rather than a human. A mandatory pop-up reminder is required when continuous interaction exceeds two hours. When users exhibit extreme emotions, providers must generate comforting content, and when self-harm or suicide is indicated, they must take assistance measures and notify guardians or emergency contacts.

Protection of Vulnerable Groups

The measures impose particularly stringent requirements for minors. Virtual intimate relationships are strictly prohibited for all minors. Services to children under 14 require guardian consent, and providers must implement a mandatory “minor mode” with time limits, content filtering, and guardian controls. Technical measures must identify minors and automatically switch to the minor mode.

For elderly users, providers must offer enhanced guidance on safe usage, prominent safety warnings, and timely responses to inquiries and assistance requests, as People’s Daily detailed.

Data Governance and Security

The regulations establish strict requirements for both training data and user interaction data. Training data must come from lawful sources and comply with core socialist values. Interaction data must be encrypted with access controls, cannot be provided to third parties without consent, and sensitive personal information in interaction data cannot be used for model training without separate user consent. Users have rights to copy and delete their interaction data.

Security assessments are required when launching services, adding new functions, using new technologies, exceeding 1 million registered users or 100,000 monthly active users, or when national security risks arise.

Penalties and Enforcement

The regulations establish a tiered penalty framework. Initial violations may result in warnings, public criticism, and orders to rectify. Escalated penalties include service suspension and fines of RMB 10,000 to RMB 100,000. Aggravated cases involving actual harm to life or health carry additional fines of RMB 100,000 to RMB 200,000. Violations may also trigger penalties under the Personal Information Protection Law, which allows fines up to RMB 50 million or 5% of annual turnover, as law firm Bird & Bird noted in its analysis.

Broader Significance

The measures represent the fourth pillar in China’s progressive AI regulatory chain, following the Algorithmic Recommendation Provisions (2022), Deep Synthesis Provisions (2023), and Generative AI Measures (2023). International law firm Hogan Lovells described the framework as “the first set of regulatory rules in China specifically targeting AI-driven emotional interaction,” marking a shift from technology-focused regulation toward scenario-based governance.

Chen Liang, Dean of the AI Law School at Southwest University of Political Science and Law, told Xinhua that the measures “bring not only compliance obligations but also an important opportunity to force industry transformation and upgrading,” urging enterprises to shift from pursuing immersive experiences and high user stickiness to exploring healthy, beneficial human-computer interaction.

What to Watch

Implementation remains an open question. Experts including Fan Kefeng, Vice President of the China Electronics Standardization Institute, have called for further refinement of regulatory measures, strengthened cross-departmental coordination, and the development of supporting standards. The interaction between these rules and evolving global AI governance frameworks, including the EU AI Act, will be closely watched by multinational companies and policymakers worldwide.

As the first dedicated regulatory framework globally for AI emotional interaction services, China’s approach is likely to influence regulatory discussions in other jurisdictions grappling with the rapid proliferation of AI companions and emotionally responsive AI systems.