China's synthetic-content labeling rules cover visible disclosure, file metadata and platform verification/notice processes. AI short-drama compliance therefore cannot be reduced to one opening notice. Disclosure needs continuity from asset generation to master export and platform upload and must explain what happened after transcode or editing.
The rules combine visible and embedded disclosure
The official measures require visible and embedded labels in applicable cases and define verification, notice and user-declaration processes for distribution platforms.
The public rules establish the framework, while applicability, display and responsibility require project-specific review by service type, content form and release route.
Disclosure passes through several fragile steps
A generation tool may write metadata, an editor may strip it, an export template may rebuild it and platform transcode may change the file. Visible disclosure can also be cropped from secondary edits.
Teams need hashes, disclosure-check results and owners at each stage so the final platform state resolves back to the master rather than one upload screenshot.
Run disclosure checks before every export
Review should cover applicable opening/player notices, file metadata, generation tool and version, asset origins, platform declaration fields and final-master hash.
Language and platform variants need separate acceptance; a passing Chinese master does not prove that an English subtitle cut, short edit or platform transcode preserves the same disclosure.
What remains unknown: This is industry analysis of public rules, not legal or regulatory advice. Projects should verify requirements for their actual tools, content, entity and platform with qualified professionals and relevant authorities.