China Deep Synthesis Provisions: CAC Labelling for AI Content
Definition: China's Deep Synthesis Provisions are the CAC rules requiring both explicit (visible) and implicit (metadata) labelling of AI-generated and synthetically altered content, in force since 10 January 2023 and reinforced by the GB 45438-2025 national standard.
TL;DR: China requires dual labelling: an explicit visible label and an implicit machine-readable mark. The Deep Synthesis Provisions took effect 10 January 2023, and GB 45438-2025 set the national labelling standard effective September 2025. Non-compliance risks service suspension, fines, and criminal liability.
Explicit and implicit, both required
The explicit label is what a viewer sees: a visible marker that content is AI-generated. The implicit label is machine-readable: metadata embedded in the file. A signed manifest plus a watermark satisfies the implicit side, while the visible label is a UI obligation on the service. China was early to require both, and GB 45438-2025 codified the format.
What the standard adds
GB 45438-2025 specifies how the labels must be applied across modalities, giving providers a concrete target rather than a principle. For platforms serving Chinese users, the metadata layer maps to the same provenance stack used for EU and California compliance, with the visible-label step added.
Operating across regimes
A provider marking output for Article 50 and SB 942 already produces the implicit label China expects. The incremental work is the explicit visible label and conformance with the GB standard's specifics, layered on the same signing pipeline.
The incident behind this
China's dual-labelling regime predates the Western wave: the Deep Synthesis Provisions took effect on 10 January 2023, making it the first major jurisdiction to require both visible and machine-readable marks on synthetic media.
Regulatory mapping
| Regime | Effective | Bite | Why it applies |
|---|---|---|---|
| CAC Deep Synthesis Provisions | 10 Jan 2023 | Suspension, fines, criminal liability | Explicit and implicit labelling |
| GB 45438-2025 | Sep 2025 | Standard conformance | National labelling format |
FAQ
Does a watermark satisfy China's rules?
It satisfies the implicit, machine-readable side. China also requires an explicit visible label, which is a UI obligation on the service.
Where Original Pictures stands today
Original Pictures ships three things today: a Sign API, a Verify API, and the SDKs that wrap them. One POST /v1/sign attaches a C2PA-format manifest, an invisible TrustMark watermark, and an OpenTimestamps anchor. The open-source verifier checks any of it without calling us.
Two things are on the near roadmap, and we name them as roadmap, not as shipped: C2PA Conformance Program recognition (target Q3 2026, until then our manifests use the published C2PA v2.2 format and any C2PA-aware validator can read them, but third-party validators will show our signer as not-yet-listed), and a consumer capture app (Q3 2026). We do not sell a capture SDK, and we do not claim Trust-List membership we do not yet hold.
Bottom line: China requires both a visible label and a machine-readable mark. Reuse your Article 50 marking pipeline for the implicit layer and add the explicit visible label to serve Chinese users.
Related
Original Pictures is progressing through the C2PA Conformance Program; our signing certificate is not yet on the official C2PA Trust List. Target: Q3 2026. We will not describe ourselves as "C2PA-certified" until it is true.
Original Pictures provides content-provenance infrastructure. It does not by itself constitute legal compliance with the EU AI Act or any other regime; compliance depends on how you deploy it, your disclosures, and your governance. Figures are drawn from public reporting, verify against primary sources before citing in regulated materials. Nothing here is legal advice.
Last verified 2026-05-25. Author: Mahdi Kazempour, Founder, Original Pictures.