Regulations · China
AI REGULATION

China

ENACTED

China has enacted multiple targeted AI regulations covering algorithms, deepfakes, and generative AI, combining innovation promotion with strict content and security controls.

KEY LEGISLATION 5 laws
Interim Measures for the Management of Generative AI Services ENACTED
2023-08
Requires security assessments for generative AI services, mandates training data legality, and enforces content alignment with socialist core values
Provisions on the Management of Algorithmic Recommendations ENACTED
2022-03
Regulates recommendation algorithms, requires transparency and user control over algorithmic feeds
Deep Synthesis Provisions (Deepfake Regulations) ENACTED
2023-01
Requires labeling of AI-generated content and prohibits deepfakes without consent
AI Safety Governance Framework ENACTED
2024-09
TC260 framework establishing AI safety requirements and risk classification
Artificial Intelligence Law (comprehensive) DRAFT
2024-12
Proposed comprehensive AI law consolidating existing regulations into unified framework
ANALYSIS

Regulatory Landscape

China has taken a uniquely rapid and targeted approach to AI regulation, enacting multiple specific rules addressing individual AI applications rather than waiting for comprehensive legislation. Since 2022, China has implemented regulations covering algorithmic recommendations, deepfakes, and generative AI services—moving faster than any other major economy.

China’s regulatory philosophy balances two priorities: maintaining global AI competitiveness (with explicit goals to lead in AI by 2030) and ensuring AI development aligns with state interests, social stability, and content control requirements.

Key Laws

The Generative AI measures, effective August 2023, require providers to conduct security assessments before launching services, ensure training data is legally obtained, and maintain content aligned with “socialist core values.” This last requirement makes China’s AI regulation uniquely tied to political ideology.

The algorithm recommendation rules were among the world’s first, requiring platforms to offer users the option to disable algorithmic recommendations and prohibiting algorithms that promote addiction or discriminate on pricing.

The deepfake provisions require clear labeling of all AI-generated synthetic content and prohibit creating deepfakes of individuals without their consent.

Enforcement

The Cyberspace Administration of China (CAC) serves as the primary enforcer, with authority to require security assessments, conduct inspections, and impose penalties. Enforcement has been active—multiple companies have been required to modify their AI services following regulatory reviews. The algorithm registry requires companies to file details of their recommendation systems with authorities.

Business Impact

Foreign companies offering AI services in China face significant compliance requirements including data localization, content filtering, and security assessments. The regulatory environment favors domestic champions while creating barriers for international competitors. However, China’s approach also provides regulatory clarity that allows compliant companies to operate with reasonable certainty about requirements.