From August 2, 2026, any AI-generated or manipulated content that resembles existing persons, places, or events must be disclosed as artificial. This requires both human-visible labels and machine-readable metadata such as C2PA provenance standards or digital watermarks.
Deepfake Labelling Requirements: What You Must Implement by August 2026
What happened
Article 50(4) of the EU AI Act targets synthetic media — commonly known as deepfakes. Any person who uses AI to generate or manipulate image, audio, or video content that "appreciably resembles" existing persons, places, or events must disclose that the content is artificially generated or manipulated.
Article 50(2) separately requires that providers of AI systems generating synthetic content must ensure the output is marked in a machine-readable format that allows detection. This means both the creator and the technology provider share compliance obligations.
What it means for your business
Three distinct obligations apply from August 2, 2026:
| Who | Obligation | How |
|---|---|---|
| AI system providers | Mark AI-generated content as machine-readable | C2PA metadata, watermarking, steganography |
| Content deployers | Disclose AI origin of deepfake content | Visible labels, captions, disclaimers |
| Content publishers | Not remove or disable provenance markers | Preserve C2PA metadata in publishing workflows |
The C2PA (Coalition for Content Provenance and Authenticity) standard has emerged as the leading technical approach. Major platforms including Google, Microsoft, Adobe, and Meta already support C2PA. The GPAI Code of Practice, published July 10, 2025, references C2PA as a recommended implementation.
What to do now
- Audit your AI content pipeline. Identify every point where AI generates or modifies images, audio, video, or text
- Implement C2PA metadata. For image/video generation: integrate the C2PA SDK into your output pipeline. For text: embed provenance metadata
- Add visible disclosure. AI-generated content shared publicly must carry visible labels — not just machine-readable metadata
- Protect provenance data. Ensure your CMS, social media tools, and distribution channels do not strip C2PA metadata during upload or processing
- Train your team. Everyone who publishes AI-generated content must understand the labelling requirements
Check your AI content compliance readiness.
Free AI Act Readiness Check Covers deepfakes, synthetic content, and transparencyThis article is for informational purposes only and does not constitute legal advice. Regulatory requirements change frequently — verify current rules with official sources. Published June 17, 2026 by Sawai Gyoseishoshi Office, Hiroshima, Japan.