Countries / China

China

China regulates AI through binding CAC rules built on mandatory algorithm filing and security assessment, plus mandatory content-labelling rules and standards, and has put a comprehensive AI law on its 2026 legislative plans.

Registry-led binding rulesApproach
9Rules tracked
7Binding and in force
0Upcoming deadlines

Latest developments in China

See all

In force (8)

  • Generative AI Measures
    RegulationCyberspace Administration of China (CAC) with NDRC, MOE, MOST, MIIT, MPS and NRTABinding

    Applies to providers of generative AI services (text, image, audio, video) to the public in mainland China. Requires lawful training data that respects IP, content controls, labelling of generated content, user protections and, for services with 'public opinion attributes or social mobilisation capacity', a security assessment plus algorithm filing.

    In forceEffective 15 Aug 2023
  • Deep Synthesis Provisions
    RegulationCyberspace Administration of China (CAC), MIIT, MPSBinding

    Regulates providers and technical supporters of 'deep synthesis' (deepfake-type) services generating or editing text, voice, images, video and virtual scenes. Requires real-name user verification, conspicuous labelling of synthetic content that could confuse the public, consent for editing biometric features, and algorithm filing.

    In forceEffective 10 Jan 2023
  • Algorithm Recommendation Provisions
    RegulationCyberspace Administration of China (CAC), MIIT, MPS, SAMRBinding

    Governs recommendation, ranking, generation/synthesis, dispatch and decision algorithms used by internet services in China. Requires algorithm filing with CAC for services with public-opinion attributes, user options to turn off personalisation, protections for minors, the elderly, gig workers and consumers (e.g. no discriminatory pricing).

    In forceEffective 1 Mar 2022
  • AIGC Labelling Measures
    RegulationCyberspace Administration of China (CAC), MIIT, MPS, NRTABinding

    Requires service providers to add explicit (visible/audible) labels and implicit labels (metadata) to AI-generated text, images, audio, video and virtual scenes, and requires platforms distributing content to detect and flag AI content. App stores must check labelling compliance.

    In forceEffective 1 Sep 2025
  • GB 45438-2025 labelling standard
    StandardState Administration for Market Regulation (SAMR) and Standardization Administration of China (SAC); proposed by the Office of the Central Cyberspace Affairs CommissionBinding

    Mandatory national standard released alongside the CAC labelling measures, specifying how explicit labels and implicit (metadata) labels must be implemented for AI-generated text, images, audio, video and virtual scenes.

    In forceEffective 1 Sep 2025
  • Cybersecurity Law amendment (AI)
    LawStanding Committee of the National People's CongressBinding

    Amends China's Cybersecurity Law to add a dedicated article (Art. 20) supporting AI research, infrastructure and training-data resources while requiring improved AI ethics norms, risk monitoring/assessment and safety supervision. The amendment also raises fines, up to RMB 10 million for especially serious harm.

    In forceEffective 1 Jan 2026
  • Anthropomorphic (companion) AI Measures
    RegulationCyberspace Administration of China (CAC), NDRC, MIIT, MPS, SAMRBinding

    Covers AI services offered to the public in China that simulate human personality, thinking and communication style for sustained emotional interaction (AI companions, virtual partners). Bars 'virtual relative' or 'virtual partner' services for minors and requires a minors' mode and guardian consent for under-14s, reminders after every 2 hours of continuous use, easy exit, protections for elderly users, algorithm filing, and a security assessment when launching or when reaching 1 million registered or 100,000 monthly active users.

    In forceEffective 15 Jul 2026
  • AI Agents Opinions
    GuidelineCyberspace Administration of China (CAC), NDRC, MIITNon-binding

    Policy document implementing the State Council 'AI+' action opinions for AI agents (systems with autonomous perception, memory, decision and execution). It sets principles (secure and controllable, orderly, innovation-driven), calls for technical standards and protocols, product rules and safety safeguards, industry self-regulation, and lists 19 priority application scenarios.

    In forceAdopted 8 May 2026

Proposed or in consultation (1)

  • Planned AI law
    BillState Council / NPC Standing CommitteeBinding

    The State Council 2026 legislative work plan (Guobanfa [2026] No. 14, dated 8 May 2026) commits to 'accelerating comprehensive legislation on the healthy development of AI' and to legislation on data, compute, algorithms and key AI applications. The NPCSC 2026 plan lists AI healthy-development legislation as a research-and-drafting item to be scheduled for review as appropriate.

    ProposedProposed 8 May 2026

What the rules require

Obligation types found across China's instruments. Filled dot: imposed by a binding instrument. Ring: guidance only.

Transparency to usersAlgorithm Recommendation Provisions, AIGC Labelling Measures, Anthropomorphic (companion) AI Measures
Labelling AI contentGenerative AI Measures, Deep Synthesis Provisions, AIGC Labelling Measures, GB 45438-2025 labelling standard
Registration or filingGenerative AI Measures, Deep Synthesis Provisions, Algorithm Recommendation Provisions, Anthropomorphic (companion) AI Measures
Human oversightAI Agents Opinions
Data governanceGenerative AI Measures
Bias and non-discriminationAlgorithm Recommendation Provisions
Copyright and training dataGenerative AI Measures
Security testingGenerative AI Measures, Deep Synthesis Provisions, Cybersecurity Law amendment (AI), Anthropomorphic (companion) AI Measures, AI Agents Opinions
Consumer rightsGenerative AI Measures, Deep Synthesis Provisions, Algorithm Recommendation Provisions, Anthropomorphic (companion) AI Measures
PenaltiesCybersecurity Law amendment (AI), Anthropomorphic (companion) AI Measures

Timeline