Administrative Provisions on Deep Synthesis Internet Information Services

Specifically addresses deepfake technology and other deep synthesis services, requiring clear labeling of AI-generated content and obtaining consent for using individuals' personal information in deep synthesis. Establishes requirements for identity verification and content moderation.

JurisdictionChina
CategoryAI Regulations
StatusActive
Latest development

Analysis

Key Requirements

The Administrative Provisions on Deep Synthesis Internet Information Services establish several key requirements for providers of deep synthesis services:

  • Labeling of synthetic content: When providing services that may generate or significantly alter information content, providers must add prominent labels to indicate the use of deep synthesis.
  • Consent for biometric information: Providers must obtain separate consent from individuals before editing their biometric information, such as facial features or voice prints.

Compliance Challenges

Organizations face several challenges in complying with these regulations:

Implementation Best Practices

To effectively implement these regulations, organizations should:

Recent Updates

As of the current date, there have been no significant recent updates to the Administrative Provisions on Deep Synthesis Internet Information Services. The regulation came into effect on January 10, 2023.

Related Regulations

The Administrative Provisions on Deep Synthesis Internet Information Services interact with several other Chinese regulations:

These laws provide the broader framework for data protection and cybersecurity within which the deep synthesis regulations operate.

Industry Impact

The regulations have had significant impacts across various industries:

Sources

Recent developments

  • — The Cyberspace Administration of China (CAC) released the fourteenth batch of record information for deep synthesis service algorithms, with 680 newly filed algorithms, reflecting ongoing regulatory oversight and compliance with the Administrative Provisions on Deep Synthesis Internet Information Services[11]. (source)
  • — Recent commentary highlights that China's deep synthesis regulation is influencing technology contracting, requiring companies to adapt compliance strategies and contractual terms to meet evolving regulatory demands[6]. (source)
  • — A Hong Kong listing document for a Chinese AI company highlights that the Deep Synthesis Provisions, together with the 2025 Identification Measures, require real‑name verification, data security controls and systematic labelling of deep‑synthesis content, describing these rules as key regulatory risks for AI and content platforms.[8] (source)
  • — The same filing reports that providers with public opinion attributes or social‑mobilization capacity must complete algorithm recommendation filings under related CAC rules, showing how the Deep Synthesis Provisions now interact with broader AI and algorithm governance measures in China’s regulatory stack.[8] (source)
  • — A global AI regulatory tracker submitted to the U.S. Congress identifies China’s Deep Synthesis Provisions as a core component of its AI framework and notes their integration with subsequent Generative AI Measures and content‑identification rules, underscoring growing international attention to China’s deep synthesis regime.[9][10] (source)
  • — Four Chinese regulators issued the **Measures for the Labelling of Artificial Intelligence-Generated and Synthetic Content**, effective 2025-09-01, which operationalize and expand labelling obligations for deep synthesis content under the 2023 Administrative Provisions.[7][8] (source)
  • — Legal commentary explains that the new labelling Measures apply to internet information service providers offering AI generation/synthesis, mandate metadata-based implicit labels and visible explicit labels, and establish verification duties for content dissemination services, significantly tightening compliance expectations under China’s deep synthesis regime.[7] (source)
  • — Industry analysis notes that high‑risk services identified in the Deep Synthesis Provisions—such as smart dialogue, voice synthesis and face generation—must carry clear labels, and that providers must update user agreements to reflect new labelling and logging obligations, increasing operational and product‑design burdens.[7] (source)
  • — Commentators observe that the new labelling Measures allow certain users to request synthetic content without explicit labels if obligations are contractually clarified and logs retained, raising practical questions for industry on balancing user experience, traceability and compliance under the Deep Synthesis Provisions.[7] (source)
  • — The article reports that service providers must deploy technical measures to verify implicit labels in metadata and add explicit labels where needed, which is prompting companies to invest in detection, content‑management and audit tooling to remain compliant with deep synthesis‑related identification duties.[7] (source)
  • — Legal experts frame the Measures as a “next step” after the Deep Synthesis Provisions, signaling China’s shift from high‑level obligations to detailed implementation rules and suggesting that further refinements to AI content governance can be expected as industry practices and enforcement mature.[7] (source)
  • — The CAC and three other agencies released the **Measures for the Identification of AI-Generated and Synthesized Content**, requiring explicit and implicit identification of AI-generated text, audio, images, video and virtual scenes that fall within the scope of the Deep Synthesis Provisions.[8][7] (source)

Read the full guide to Administrative Provisions on Deep Synthesis Internet Information Services

Related regulations

Put it into practice

Browse all regulations · Compliance deadlines · Latest updates