AI Regulation in China
China has enacted the most prescriptive and rapidly evolving AI regulatory framework in the Asia-Pacific region. Three major regulations apply to AI services accessible to Chinese users: the Algorithm Recommendation Regulations (2022) covering content recommendation systems; the Deep Synthesis Regulations (2022) requiring labeling of AI-generated images, video, and audio; and the Generative AI Regulations (2023) imposing security assessments and real-name registration on providers of publicly accessible generative AI services. All are administered by the Cyberspace Administration of China (CAC).
The Generative AI Regulations are the most operationally significant for international organizations. Services offering generative AI capabilities to users in China must complete a security assessment and register with the CAC before public launch. Training data must comply with labeling obligations, and outputs must carry disclosure notices. The regulations apply to foreign companies serving Chinese users, not just domestic operators.
China's regulatory model reflects priorities distinct from Western frameworks: content control, algorithmic sovereignty, national security alignment, and Communist Party oversight sit alongside technical safety requirements. The compliance stack operates largely independently of the EU AI Act or NIST AI RMF, requiring organizations with Chinese market presence to maintain a separate compliance program tailored to CAC requirements and the political dimensions of AI governance in China.
Key themes
- 1.Generative AI registration and security assessments (CAC)
- 2.Deep synthesis labeling obligations for AI-generated media
- 3.Algorithm recommendation filing requirements
- 4.National security and content control obligations
Regulatory frameworks and guidance(8)
China AI Standardization White Paper
Chinese standards authorities issued this non-binding policy document to map AI standards work. It identifies priorities and outlines planned national and international standardization efforts.
China Algorithm Recommendation Regulations
China's dedicated regulation on recommendation algorithms covers internet services operating in China or targeting Chinese users. Providers must meet transparency, user-control, and content-moderation duties.
China Deep Synthesis Regulations
China’s deep synthesis provisions regulate service providers and users operating in or targeting China. They cover AI-generated or manipulated text, images, audio, video, and virtual humans. Duties include labeling, registration, security assessments, and content moderation.
China Draft AI Law
China is developing a proposed national AI law. It would establish legal duties for AI development, deployment, and governance across sectors.
China's Interim Measures for the Management of Generative Artificial Intelligence Services
China’s Interim Measures for Generative AI Services cover providers serving the Chinese public. They impose duties for training data governance, content safety, algorithmic transparency, and registration.
Implementation Opinions on the Standardized Application and Innovative Development of Intelligent Agents
Three Chinese regulators released these Implementation Opinions on 8 May 2026, China's first policy framework dedicated to AI agents. The document is a policy framework rather than a binding regulation. It sets principles on user control and agent authority, with filing, compliance testing, and recall expected in sensitive sectors.
China's Interim Measures for Artificial Intelligence Anthropomorphic Interactive Services
China’s Interim Measures for Artificial Intelligence Anthropomorphic Interactive Services govern AI that simulates human interaction. Covered services include chatbots and task-performing AI products operating within Chinese jurisdiction. Providers and deployers must define authorization boundaries (limits on what the AI may do), use tiered risk approvals, and meet registration duties before public availability.
China's Measures for Labelling AI-Generated and Synthetic Content
China’s labeling measures cover AI-generated and synthetic content distributed to Chinese users. Platforms, developers, and enterprises must label covered text, images, audio, and video. Required mechanisms include audio Morse codes, encrypted hidden file data, and labels that work in virtual reality.