detect·deepfakesby Resemble AI
Deepfake law · China

Deepfake Law in China

China has the strictest deepfake regulation of any major economy — mandatory watermarking of AI-generated content since 2023, consent requirements for subject depiction, and platform obligations enforced by the CAC.

Status
enacted
Jurisdiction
China
Effective
Jan 2023
Statute
Regulations on the Administration of Deep Synthesis of Internet Information Services 2023
Mandatory watermarkingConsent requirementsPlatform liabilityAI service provider licensing
Updated Apr 16, 2026 · 1 min read

China implemented the world's first comprehensive AI-generated content regulation in January 2023. Under the "Deep Synthesis Regulations" and subsequent rules, AI-generated content must be watermarked, subjects must consent to depiction, and platforms are liable for hosted content.

Key provisions

Deep Synthesis Regulations (effective January 2023). Require:

  • Mandatory watermarking (visible or machine-readable) on all AI-generated synthetic content distributed in China.
  • Consent from any real person whose likeness or voice is used in synthetic content.
  • Platform review of AI-generated content before distribution.
  • Service provider licensing — operators of deep-synthesis services must register with the CAC.
  • Real-name verification for users of synthetic-media tools.

Interim Measures for Generative AI Services (August 2023). Supplemented the earlier regulations for general-purpose generative AI (ChatGPT-type services). Requires content-moderation aligned with Chinese laws (including political speech restrictions), algorithm filing with the CAC, and socialist-values alignment for training data.

Personal Information Protection Law (PIPL). Biometric data provisions apply to voice and facial features in synthetic content.

CAC enforcement

The Cyberspace Administration of China (CAC) is the primary enforcer. Enforcement activity through 2024–2025 has included:

  • Suspension of multiple deep-synthesis service providers for non-compliance.
  • Fines on platforms hosting unlabeled AI-generated content.
  • Takedown orders and account suspensions.

Practical implications

For organizations operating in China or targeting Chinese users:

  • AI service providers: licensing is mandatory. Operating a deep-synthesis service in China without CAC registration is not legally viable.
  • Platforms: pre-distribution review obligations require substantial content-moderation infrastructure.
  • Foreign operators: the combination of data-localization requirements, licensing, and content-moderation obligations makes China challenging for many foreign AI providers.

The Chinese framework is the global precedent for mandatory watermarking and is frequently cited in policy debates elsewhere.

Sources