Measures for Labeling AI-Generated and Synthetic Content

Summary

Sets rules for labelling AI-generated and synthesised content (text, images, audio, video, virtual scenes) in 14 articles. Providers of services covered by Art. 17(1) of the Deep Synthesis Provisions must add explicit labels that users can clearly perceive, and providers must add implicit labels in file metadata, with digital watermarks encouraged. Platforms that spread content must check metadata and add prominent notices to content that is labelled, declared or suspected as generated; app distribution platforms must verify providers' labelling materials; and maliciously deleting, tampering with, forging or concealing labels is prohibited.

Language

English translation by China Law Translate, not by a public body Open it. The Chinese text is authoritative.

Safety topics

Status notes

Signed 7 Mar 2025 and published 14 Mar 2025; effective 1 Sept 2025 (Art. 14). The Measures implement the labelling duties in the algorithmic recommendation, deep synthesis and generative AI rules (Arts. 1 and 2). Art. 11 also requires labelling to meet mandatory national standards; the mandatory national standard GB 45438-2025 on labelling methods (CN-08) took effect on the same day. Users may obtain content without explicit labels only after a user agreement sets out their labelling duties and responsibility for use, and providers keep logs of such provision for at least six months (Art. 9); implicit labels still apply.

Sources

Open the official source https://www.cac.gov.cn/2025-03/14/c_1743654684782215.htm

Secondary source: https://www.gov.cn/zhengce/zhengceku/202503/content_7014286.htm

Local copy

Verification

History

1. 2026-10-14related: [{"slug": "cn-gb-45438-2025", "id": "CN-08"}, {"slug": "cn-deep-synthesis-provis → [{"slug": "cn-gb-45438-2025", "id": "CN-08"}, {"slug": "cn-deep-synthesis-provis — two-way link to CN-20, CN-23 added in run 2026-10-14-discover-leads-2026-10-142026-10-14-discover-leads-2026-10-14 · Claude Code (claude-opus-5-5)
2. 2026-10-14english_version: – → {"kind": "third-party translation", "url": "https://www.chinalawtranslate.com/en — Return round (14 Oct 2026): set to China Law Translate's translation, now stored ('By China Law Translate on 2025/03/14', title 'Measures for Labeling of AI-Generated Synthetic Content'). The findings and stored files show no English version by a public body.2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
3. 2026-10-14notes: Signed 7 Mar 2025 and published 14 Mar 2025, together with the mandatory nationa → Signed 7 Mar 2025 and published 14 Mar 2025; effective 1 Sept 2025 (Art. 14). Th — 'Requests for unlabelled output' corrected: Art. 9 allows content without explicit labels (不含显式标识), and implicit labels (Art. 5) still apply. 'Published together with GB 45438-2025' replaced by the Art. 11 reference and the shared effective date, because the stored text does not name the standard. Dropped the years of the three earlier rules, which the stored text does not give.2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
4. 2026-10-14summary: Requires explicit (user-visible) labels and implicit labels (file metadata, with → Sets rules for labelling AI-generated and synthesised content (text, images, aud — The explicit-label duty is now tied, as in Art. 4, to services under Art. 17(1) of the Deep Synthesis Provisions; 'user-visible' became 'that users can clearly perceive' (Art. 3, 明显感知). Platform duties now cover content that is labelled, declared or suspected as generated (Art. 6, items 1 to 3). 'App stores' became 'app distribution platforms' (互联网应用程序分发平台), which verify labelling materials (Art. 7). 'Removing, altering' became 'deleting, tampering with' (删除、篡改, Art. 10).2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)

Related records

GB 45438-2025 (AI content labeling standard) China · In force

Deep Synthesis Provisions China · In force

Generative AI Measures China · In force

Algorithmic Recommendation Provisions China · In force

TC260-PG-20268A (AI application security, general provisions) China · Published (non-binding, current)

TC260-PG-202611A (AI application security, radio, TV and network audiovisual) China · Published (non-binding, current)

Citation

Cyberspace Administration of China; Ministry of Industry and Information Technology; Ministry of Public Security; National Radio and Television Administration. Measures for Labeling AI-Generated and Synthetic Content [国信办通字〔2025〕2号]. 2025-03-07. https://www.cac.gov.cn/2025-03/14/c_1743654684782215.htm. Status: In force; verified 2026-10-14 (level A). Local copy archive/CN-09/2026-10-11/official.html retrieved 2026-10-11 14:00:13 UTC, SHA-256 3686321b6f2fca2e245e210209dff4af75b5adc3be36d912695d28b5a681f642. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record CN-09.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.