6 binding regulations · Active

China enforces multiple AI laws — and your AI products sold in China must comply.

China has the most active AI enforcement regime in the world, with multiple binding regulations already in effect: the Generative AI Services Management Measures, Algorithmic Recommendation Regulations, Deep Synthesis Regulations, and synthetic content identification rules. Any company selling AI-powered products or services in the Chinese market faces immediate compliance obligations including consent, data quality, content labeling, user rights, and complaint handling. AI Asset Assurance provides independent evaluation documentation that supports compliance with China's AI regulatory requirements.

Request evaluation

REGULATION STATUS

GenAI MeasuresEffective Aug 2023

Algorithm RulesEffective Mar 2022

Deep SynthesisEffective Jan 2023

Synthetic ID rulesEffective Sep 2025

AI Asset Assurance coverageBias & fairness evaluation

China's AI regulatory stack

Four regulations, all enforcing now

China's approach is regulation-by-technology-type rather than risk-based classification. Each AI technology has its own regulatory framework with the Cyberspace Administration of China (CAC) as the primary enforcer.

Generative AI Services Management Measures

Effective August 15, 2023

Applies to GenAI services offered to the public in China. Requires: training data legality and quality, content safety measures, user consent for data collection, complaint mechanisms, and algorithm filing with the CAC. Providers must ensure GenAI does not generate content that undermines social stability.

AI Asset Assurance: Training data quality and demographic representativeness evaluation directly supports the data quality requirements. Bias evaluation ensures the system doesn't produce discriminatory outputs across Chinese regulatory categories.

Algorithmic Recommendation Regulations

Effective March 1, 2022

Regulates algorithmic recommendation systems including content feeds, search rankings, and product recommendations. Users must be able to opt out. Algorithms must not create filter bubbles, price discrimination, or addictive patterns. Algorithm filing with CAC required.

AI Asset Assurance: Algorithmic fairness evaluation identifies whether recommendation systems produce differential treatment by user demographics — including the price discrimination and demographic steering that the regulations specifically prohibit.

Deep Synthesis & Synthetic Content Rules

Effective January 10, 2023 + September 1, 2025

Regulates deepfakes and AI-generated content. Requires watermarking and labeling of synthetic content. Providers must verify user identities, maintain logs, and implement content moderation. The 2025 update adds specific identification requirements for AI-generated images, video, and audio.

AI Asset Assurance: While AI Asset Assurance focuses on bias and fairness rather than content moderation, the evaluation documents the AI system's decision-making process — supporting the transparency and documentation requirements that underpin all Chinese AI regulations.

Cross-border compliance

Multiple regulations

US companies operating in China — or whose AI products are accessible to Chinese users — face compliance obligations under all applicable regulations. Joint ventures, technology licensing, and SaaS platforms serving Chinese users are all in scope. Non-compliance risks include service suspension, fines, and market access restrictions.

AI Asset Assurance: Cross-jurisdictional evaluation provides documentation that supports compliance with Chinese, EU, and US requirements simultaneously. Companies entering or operating in the Chinese market can demonstrate proactive compliance with independent evaluation.