AI-Generated Synthetic Content Labelling Measures

On 14 March 2025 the Cyberspace Administration of China published a notice circulating the Measures, jointly formulated with the Ministry of Industry and Information Technology, the Ministry of Public Security and the National Radio and Television Administration. The notice number is 国信办通字〔2025〕2号; the measures themselves are dated 7 March 2025.

Article 1 states the purpose: promote healthy AI development, regulate labelling of AI-generated synthetic content, protect the lawful rights of citizens, legal persons and other organisations, and safeguard the public interest, pursuant to the Cybersecurity Law, algorithm-recommendation rules, deep-synthesis rules and the interim generative-AI service measures.

Article 14 sets the effective date: 1 September 2025.

Article 2 limits scope to network information service providers that fall under specified algorithm-recommendation, deep-synthesis or generative-AI service rules when they conduct AI-generated synthetic content labelling.

Article 3 defines AI-generated synthetic content as text, images, audio, video, virtual scenes and other information generated or synthesised with AI. Labelling has two forms: explicit labels (visible or audible marks users can clearly perceive) and implicit labels (technical marks in file data that users do not clearly perceive).

Article 4 requires explicit labels for deep-synthesis services covered by Article 17(1) of the deep-synthesis rules, with modality-specific placement rules for text, audio, images, video and virtual scenes, and a duty to keep explicit labels in files that users download, copy or export.

Article 5 requires implicit labels in file metadata under deep-synthesis Article 16, covering generation attributes, provider name or code and content identifiers. Digital watermarks are encouraged. Metadata is defined as descriptive information embedded in the file header.

Article 6 places duties on distribution platforms: check metadata for implicit labels and add prominent notices when content is marked, user-declared, or detected as suspected synthetic; offer labelling tools; and add propagation metadata when publishing under those cases.

Articles 7–12 cover app-store verification of generative services, contract disclosures, optional provision of unmarked content after logging user responsibility for at least six months, user declaration duties, bans on malicious removal or forgery of labels, alignment with mandatory standards, and use of labelling materials in algorithm filing and security assessment.

Article 13 assigns enforcement to cyberspace, telecoms, public-security and broadcasting authorities under existing laws and rules.

• Service providers · add explicit labels for covered deep-synthesis modalities (Art. 4).
• Service providers · embed implicit labels in file metadata (Art. 5).
• Distribution platforms · verify, notify and re-label synthetic or suspected synthetic content (Art. 6).
• App stores · verify generative-service labelling materials on listing (Art. 7).
• Users · declare and label when publishing synthetic content; no malicious stripping (Arts. 9–10).

Both regimes impose provider-facing labelling and marking duties for synthetic content, but they are not the same instrument. China’s measures are administrative rules nested under prior algorithm, deep-synthesis and generative-AI regimes, with platform distribution duties spelled out in Article 6.

EU Article 50 splits provider machine-readable marking from deployer deepfake and public-interest text disclosure inside a directly applicable regulation. Cross-regime comparison is inference: shared policy instrument, different legal architecture and exceptions.

Scope turns on cross-references to the algorithm-recommendation, deep-synthesis and generative-AI interim measures. Those parent instruments define which providers fall inside Article 2.

Enforcement practice after 1 September 2025, and any later mandatory standards under Article 11, can refine operational detail beyond Articles 1–14.

The following actors are linked to this vehicle in the graph. Their presence here records institutional role on the page, not an independent finding that each actor has completed a compliance duty.

• Cyberspace Administration of China (government): China’s central internet regulator and a lead authority for algorithm, synthetic-content and generative-AI governance.

Each block below is an explicit evidence row: a claim that points to one supporting passage. Read the quote and locator before treating the paraphrase as settled.

• China’s synthetic-content labelling measures entered into force. The provision fixes when covered providers and platforms became subject to the measures. Operational note: It supplies the legal date needed to distinguish adoption of the rules from their application.

Dated events attached to this vehicle. Each entry is a graph event with its own update page when present.

• 2025-03-14 · publication · China publishes synthetic-content labelling measures. Publication established the final policy text and a future effective date, giving covered providers time to prepare. Consequence recorded: Created a national labelling regime for covered synthetic media.
• 2025-09-01 · effective-date · China’s synthetic-content labelling measures take effect. The effective date moved the record from adopted text to an operative provider and platform regime. Consequence recorded: Converted the labelling policy into an enforceable administrative measure.

A vehicle is an instrument. Asks are the outcomes the instrument is treated as advancing in the AI Tribune graph.

• Require machine-readable marking of AI-generated or altered content Status: in-force. Require covered providers to attach detectable technical provenance markers to synthetic or materially altered content.
• Require visible disclosure for deepfakes and public-interest synthetic media Status: in-force. Require people encountering covered synthetic media to receive a clear, visible disclosure rather than relying only on hidden metadata.

These documents are the originals behind the evidence rows. Prefer the document text over secondary paraphrase when the two diverge.

• AI-Generated Synthetic Content Labelling Measures (Administrative measure, 2025-03-14). China’s rules for explicit and implicit labels on covered AI-generated synthetic content.

The following rows are labeled inference in the evidence graph. They combine more than one passage and must not be read as single-source facts.

• EU and Chinese rules converge on provider-side labelling: Both regimes impose provider-facing duties, although their legal scope and enforcement structure differ. Limit: This is cross-source analysis. It should not be read as a claim that terminology, exceptions or enforcement powers are identical.