Glossary · China
Deep synthesis provisions (China)
Also known as: Provisions on the Administration of Deep Synthesis Internet Information Services, deepfake rules, deep synthesis
The Provisions on the Administration of Deep Synthesis of Internet-based Information Services are Chinese regulations in effect since January 10, 2023 that govern technologies generating or editing text, images, audio, video and virtual scenes with deep learning. They require providers to verify users’ identities, label synthetic content, obtain consent for editing biometric features and file their algorithms.
- AI regulation
- China
In one sentence
China’s deep synthesis provisions (since January 2023) regulate deepfake technology: identity checks, labels, consent for face and voice editing.
Example
A face-swapping app in China verifies its users’ identities, obtains consent from the person whose face is used and adds a visible label to every generated video.
How it applies
- Duties: Verifying the real identity of users; labeling synthetic content so that it can be recognized, with prominent labels for content that may confuse or mislead the public (such as synthesized voices, face generation or face swapping); obtaining consent before editing a person's face or voice; security assessments for tools that edit biometric information; the algorithm filing; and a ban on using deep synthesis to produce or spread false news.
- Successors: The labeling duty was specified in detail by the labeling measures in 2025.
- Technical documentation: Documentation for deep synthesis tools must describe the labeling mechanism, the consent process and the controls against misuse.
Deep synthesis provisions vs. the EU AI Act
The AI Act addresses deepfakes through transparency obligations that apply from August 2026: providers mark synthetic content, deployers disclose deepfakes. China's provisions have applied since January 2023 and add identity verification and consent duties that the AI Act leaves to other laws, such as the GDPR.