Artificial Intelligence · China
AI regulation in China: laws & policy (2026)
China shaded by its artificial intelligence status
AI in China: sectoral rules.
FrameworkA layered regime of vertical rules issued mainly by the Cyberspace Administration of China (CAC) with other agencies: Interim Measures for the Management of Generative AI Services (2023), Provisions on the Administration of Deep Synthesis of Internet-based Information Services (2023), Provisions on the Administration of Algorithmic Recommendations (2022), Measures for Labeling AI-Generated Synthetic Content (2025), and the Interim Measures on Anthropomorphic Interactive AI Services (2026). A single comprehensive AI Law is drafted but not yet enacted.
China does not yet have a single, horizontal AI statute in force; instead it regulates AI through a tightly coordinated set of sectoral CAC-led rules covering generative AI, deep synthesis, recommender algorithms, and content labeling, backed by pre-launch algorithm filings and security assessments. A comprehensive AI Law has been on the NPC/State Council agenda since 2023 and is expressly listed again in the State Council's 2026 Legislative Work Plan (12 May 2026), but remains at drafting stage. Non-binding instruments — the 2017 New Generation AI Development Plan and TC260's AI Safety Governance Framework (v1.0 2024, v2.0 2025) — set the broader policy and technical direction.
Key points
The CAC-led Interim Measures for the Management of Generative AI Services took effect on 15 August 2023, requiring lawful training data, personal data protection, content moderation, transparency, and pre-launch algorithm filing and security assessment for services with 'public opinion or social mobilization' attributes.
The Provisions on Deep Synthesis (in force 10 January 2023) and the Provisions on Algorithmic Recommendations (in force 1 March 2022) impose labeling, user-control, transparency, and algorithm filing obligations on synthetic media and recommendation services.
The Measures for Labeling AI-Generated Synthetic Content, jointly issued by CAC, MIIT, MPS and NRTA, apply nationwide and require both visible ('explicit') labels and metadata-embedded ('implicit') watermarks on AI-generated text, images, audio and video; the accompanying mandatory standard GB 45438-2025 provides the technical marking method.
Foundation-model providers must file algorithms with the CAC and downstream applications must register; by 31 August 2025 the CAC had processed 538 generative-AI service filings and 263 application registrations, and providers must display model names and filing numbers on service pages.
On 10 April 2026 the CAC (with NDRC, MIIT and others) issued the Interim Measures for the Administration of Anthropomorphic Interactive AI Services and Implementation Opinions on Intelligent Agents, effective 15 July 2026, extending algorithm filing, security assessment and 'AI-generated' disclosure duties to chatbots, virtual companions and autonomous agents.
A unified 'AI Law' has been on the NPC Standing Committee agenda since 2023; it was dropped from the 2025 legislative schedule but the State Council's 2026 Legislative Work Plan (12 May 2026) again calls to 'accelerate comprehensive legislation' for AI — so a horizontal statute remains at drafting stage, not in force.
The AI Safety Governance Framework issued by TC260 (v1.0 September 2024, v2.0 September 2025) provides non-binding technical guidance, sitting on top of the 2017 State Council New Generation AI Development Plan that targets global AI leadership by 2030.
Timeline - major decisions & events
China's cabinet released a national blueprint to embed AI across six key sectors with penetration targets for 2027 and 2030, pairing development goals with pillars on policy/regulation and security. It signals AI as a central economic-growth and governance priority.
State Council (gov.cn) ↗After being listed in the 2023 and 2024 State Council legislative plans, multiple NPC deputies proposed drafting a unified AI Law in 2025, though the 2025 plan dropped it as a near-term NPC item, indicating an intended but still-pending overarching statute.
Geopolitechs (analysis) ↗The CAC and three other bodies issued mandatory explicit/implicit labeling rules (with national standard GB 45438-2025), effective 1 September 2025. They operationalize labeling duties across the algorithm, deep-synthesis, and generative-AI regimes.
Cyberspace Administration of China ↗China's national cybersecurity standards committee published a non-binding framework setting a risk taxonomy, governance principles, and technical countermeasures (including frontier/CBRN and loss-of-control risks) that feeds into binding national standards.
TC260 (National Information Security Standardization Technical Committee) ↗China's first binding generative-AI regulation, issued by the CAC with six other regulators and effective 15 August 2023, requires security assessments and algorithm filing for public-facing services with 'public-opinion' attributes. It set the core compliance regime for tools like chatbots.
Cyberspace Administration of China ↗The CAC, MIIT, and MPS adopted rules (effective 10 January 2023) requiring conspicuous labels on synthetically generated or altered content (deepfakes) and filing of deep-synthesis services. This established the foundation for later AI-content labeling rules.
Cyberspace Administration of China ↗Issued by four regulators and effective 1 March 2022, these rules govern recommendation algorithms, mandating transparency, user opt-out, protection of vulnerable groups, and algorithm filing. They were China's first binding algorithm-specific regulation and the template for later AI rules.
Cyberspace Administration of China ↗The Ministry of Science and Technology's national AI governance committee issued six fundamental ethical norms and 18 requirements covering the AI lifecycle. Though non-binding, it shaped the principles underpinning later binding regulations.
CSET (Georgetown) translation ↗The State Council's foundational strategy set staged goals to make China the world's primary AI innovation center by 2030 and explicitly called for AI standards, ethics, laws, and security, launching China's national AI governance agenda.
State Council (gov.cn) ↗China - other topics
Artificial Intelligence in other countries
Last verified 8/25/2026 · Orientation, not legal advice - verify against the primary sources linked above. Methodology & how to cite · State of Technology Regulation 2026 · Explore the full world map →