Skip to content

China · Regulation

China — Updated Enforcement Directives for Deep Synthesis and Generative AI Services

In force1 January 2025

Issued by Cyberspace Administration of China (CAC).

Summary

China’s 2025 enforcement directives for deep synthesis and generative AI services intensified implementation of existing generative AI and algorithmic governance rules. The directives focused on advanced watermarking for hyper-realistic deepfake content, requiring service providers to preserve metadata labels through editing, compression, and cross-platform sharing. The Cyberspace Administration of China also increased algorithm registration audits for recommendation and generative systems. Developers were expected to demonstrate safeguards against algorithmic manipulation and show how their systems support information diversity within China’s regulated digital environment.

Why it matters

Mandatory cryptographic watermarking; persistent metadata labels across editing, compression, and sharing; increased algorithm registration audits; controls for recommendation and generative systems; safeguards against algorithmic manipulation; information diversity requirements.

Topics

Record metadata

Country
China
Document type
Regulation
Status
In force
Publication date
1 January 2025
Last updated
27 June 2026
Language
English
Original source
View original source

Citation export

Plain citation

Cyberspace Administration of China, Regulations on the Management of Deep Synthesis and Generative AI Services: Updated Enforcement Directives, 2025.

BibTeX

@misc{china_updated_enforcement_directives_deep_synthesis_generative_ai_2025,
  title        = {China — Updated Enforcement Directives for Deep Synthesis and Generative AI Services},
  author       = {Cyberspace Administration of China (CAC)},
  year         = {2025},
  howpublished = {Global AI Governance Archive},
  url          = {https://www.cac.gov.cn/}
}

See the citation guide for style-specific examples.

Recorded cross-references

  • China — Interim Measures for the Management of Generative AI Services
  • Beijing AI Principles
  • European Union — AI Act 2025 Ban Phase-In and GPAI Guidelines

Related Entries

  • ChinaFramework1 May 2019

    Beijing AI Principles

    The Beijing AI Principles offered an early view of China’s AI governance thinking before later binding generative AI measures. Issued by the Beijing Academy of Artificial Intelligence in 2019, the principles balanced rap…

  • ChinaRegulation13 July 2023

    China — Interim Measures for the Management of Generative AI Services

    China’s Interim Measures for the Management of Generative Artificial Intelligence Services created a binding framework for public-facing generative AI services covering text, image, audio, and video outputs. The measures…

  • G20Framework1 January 2025

    G20 — Framework on Digital Public Infrastructure and Ethical AI Integration

    The G20 framework linked AI governance with Digital Public Infrastructure, reflecting a stronger Global South emphasis in 2025. It framed ethical AI integration as dependent on public digital systems, sovereign capacity,…

  • United States / ColoradoAct / Law1 January 2025

    United States — Colorado AI Act SB 24-205 Implementation and Rules

    Colorado’s AI Act implementation represented a major state-level development in United States AI governance while federal legislation remained stalled. The law established consumer protection duties for developers and de…

  • European UnionGuidance1 February 2025

    European Union — AI Act 2025 Ban Phase-In and GPAI Guidelines

    In 2025, the European Union moved from adopting the AI Act to phased enforcement. Prohibited AI practices began to apply in February, requiring organisations to remove or audit systems involving practices such as cogniti…

This archive is provided for research and informational purposes. It should not be interpreted as legal or regulatory advice.