China AI Policy Tracker: Key Regulations in English (2026)

A builder's routing guide to China AI regulations — CAC generative AI rules, synthetic media governance, algorithm requirements — each mapped to its English-accessible primary source

Thesis

China is the only major AI power with a layered, sector-specific AI regulatory stack already in force. By mid-2026, three landmark rules — the Algorithmic Recommendations Provisions (March 2022), the Deep Synthesis Provisions (January 2023), and the Generative AI Interim Measures (August 2023) — are actively enforced. The 2024 AI Safety Governance Framework adds risk classification. For global builders, the practical implications concentrate on two areas: content labeling and compliance for any product accessible to Chinese users, and data localization for training data sourced in China. This page maps each regulation to its enforcement scope, key requirements, and the fastest English-language primary source.

Decision in 20 seconds

Your concern Relevant regulation Best English source
Building a recommendation feed for China users Algorithmic Recommendations Provisions (2022) State Council English + China Law Translate
Generating synthetic media (faces, voices, video) Deep Synthesis Provisions (2023) State Council English; Xinhua summary same-day
Launching an LLM product in China Generative AI Interim Measures (2023) State Council English (full); China Law Translate (annotated)
Risk classification for your AI system AI Safety Governance Framework (2024) CAICT report (English summary); China Law Translate
Monitoring new AI rules as they appear CAC official releases (Chinese-first) Xinhua English (same-day summary); State Council English (48–72h full text)

China AI regulation timeline (2022–2026)

Regulation Effective date Regulator Primary impact English source
Provisions on the Management of Algorithmic Recommendations March 1, 2022 CAC / MIIT / MPS / SAMR Recommendation systems must allow user opt-out; prohibit filter bubbles and addictive design; real-name for news feed providers State Council English
Provisions on the Management of Deep Synthesis Internet Information Services January 10, 2023 CAC / MIIT / MPS AI-generated content labeling mandatory; consent required for face/voice synthesis; security assessments for providers State Council English
Interim Measures for the Management of Generative AI Services August 15, 2023 CAC + 6 co-regulators Content compliance for LLM products serving Chinese users; security assessment before public launch; API providers included China Law Translate (annotated)
AI Safety Governance Framework 2024 (guidance) CAICT / TC260 Risk classification tiers (general / high-risk / very high-risk); safety assessment methodology CAICT English summary; China Law Translate
Measures for the Security Assessment of Internet Information Services with Public Opinion Attributes February 2022 CAC Pre-launch security review for apps with public opinion influence, including AI-powered news aggregators State Council English; Xinhua English
Personal Information Protection Law (PIPL) November 1, 2021 CAC / SAMR Data localization and cross-border transfer restrictions affecting AI training data; consent requirements for biometric processing State Council English
Data Security Law (DSL) September 1, 2021 CAC / MIIT Classification of "important data" affecting AI training data exports; national security review for cross-border data transfers State Council English
AI Standardization Roadmap 2.0 2024 (published) TC260 / SAC National standards for AI terminology, testing, and evaluation; referenced by Generative AI Interim Measures for compliance benchmarks China Law Translate; CAICT English summaries

English-accessible policy sources

Source URL Coverage Translation lag
The State Council English english.www.gov.cn Full text of major national AI regulations; State Council AI plans and white papers 48–72h after Chinese release for major rules
Xinhua English english.news.cn Summaries of all major AI policy announcements; ministerial press conferences Same-day for major announcements; summary-level only
CAC Official cac.gov.cn Primary text of all CAC rules (deep synthesis, generative AI, algorithm regs); enforcement actions Chinese-only; State Council English covers major rules
China Law Translate chinalawtranslate.com Annotated translations of CAC, TC260, MIIT rules; technical AI standards; subscription for full archive Days to weeks; unofficial but most technically accurate

Compliance scope: who the rules apply to

The Generative AI Interim Measures (August 2023) apply to any organization providing generative AI services to users in China — including foreign companies whose products are accessible in China. This means API providers, not just consumer app developers, need to conduct security assessments before launching. The CAC has issued guidance that "public-facing" includes both mobile apps and API endpoints used by third-party developers who serve Chinese users. The practical implication: if you are building on a Chinese LLM API and distributing to Chinese users, you are likely within scope of the Interim Measures.

FAQ

Where can I track China AI policy updates in English?
The State Council English (english.www.gov.cn) and Xinhua English (english.news.cn) publish official translations of major AI regulations, typically 48–72h after the Chinese-language release. China Law Translate provides annotated unofficial translations of technical standards. RadarAI's policy tracker maps each regulation to its source and notes translation lag.
What are the key China AI regulations builders need to know?
The three most impactful for product builders: Algorithmic Recommendations Provisions (March 2022), Deep Synthesis Provisions (January 2023), and Generative AI Interim Measures (August 2023). The 2024 AI Safety Governance Framework adds risk classification. PIPL and DSL are the data layer below all three.
What is the CAC and how does it affect AI development in China?
CAC (Cyberspace Administration of China) is China's primary internet content regulator, with authority over AI-generated content, algorithm recommendations, and generative AI services. For AI builders, CAC rules require content labeling, security assessments before public launch, and data handling compliance for training data sourced in China.
How do China's AI regulations compare to EU AI Act?
EU AI Act uses a single risk-tiered framework (prohibited / high-risk / limited-risk / minimal-risk) applied market-wide from August 2026. China's approach is domain-specific: separate rules for recommendation algorithms, deepfakes, and generative AI, with enforcement divided across CAC, MIIT, and SAMR. China's rules are earlier in force and focus more on content compliance; EU focuses more on system-level conformity assessment.
Is synthetic media regulated in China?
Yes, since January 10, 2023. The Deep Synthesis Provisions require mandatory labeling for all AI-generated content (text, image, audio, video), consent for face/voice synthesis, and real-name registration for providers. Enforcement is by CAC. The regulation applies to any service accessible to users in China.
Where do I find English translations of Chinese AI regulations?
Official: State Council English (english.www.gov.cn), 48–72h lag. Summaries: Xinhua English, same-day. Technical/annotated: China Law Translate (chinalawtranslate.com). CAC primary text is Chinese-only — use State Council English for the official translated version of major CAC rules.

Companion pages in this cluster

If your question is about… Go to What's there
How China AI compares to US AI overall China AI vs US AI Ecosystem comparison across models, compute, policy, and open source
Best English sources for China AI news News Sources in English Layer-by-layer routing guide for policy, model releases, funding
Weekly digest of China AI developments Weekly China AI Digest 15-minute Monday read covering models, policy, and funding signals
Chinese LLM API access from outside China China AI API Access Guide Qwen, DeepSeek, Kimi API international availability and signup
Broad China AI context and overview China AI Overview Topic definition, cluster routing matrix, start-here guide

Quotable summary: China's AI regulatory stack is the most comprehensive in the world by enforcement date — three major rules in force since 2023, a risk classification framework from 2024, and the world's only live generative AI regulation. For English-first builders, the key routing is: State Council English for official text, China Law Translate for annotated technical detail, and Xinhua English for same-day policy signals.