Guides · Article 50 · EU AI Act
Synthetic media labeling under the EU AI Act
'Label your synthetic media' sounds like one rule, but under Article 50 of the EU AI Act it is two — one for the party that provides the generating system, and one for the party that deploys it. Understanding which side of that split you are on is the first step. This guide explains both from the regulation text. Informational only — not legal advice.
The provider side: machine-readable marking (Article 50(2))
Providers of AI systems generating synthetic audio, image, video or text must ensure the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated, using technical solutions that are effective, interoperable, robust and reliable as far as technically feasible (Article 50(2), Regulation (EU) 2024/1689). This is marking baked in at the system level — think metadata or watermarking — subject to carve-outs, including where the AI performs an assistive function for standard editing or does not substantially alter the input.
The deployer side: deepfake disclosure (Article 50(4))
Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake must disclose that the content has been artificially generated or manipulated. Where the content forms part of an evidently artistic, creative, satirical, fictional or analogous work, the obligation is limited to disclosing the existence of the generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work. A second sub-paragraph extends a disclosure duty to AI-generated text published to inform the public on matters of public interest, with its own carve-outs.
How the two fit together
In a typical pipeline, the tool vendor's marking duty and the publishing business's disclosure duty are complementary: the provider marks outputs so they are machine-detectable, and the deployer discloses to the audience when the content is a deepfake. A single business can hold both roles — building a generator (provider-side) while also publishing its outputs (deployer-side). Which paragraphs touch you is fact-specific; map each system to a role and confirm with qualified counsel. Our free scope check gives an informational starting point.
Common questions
If my AI tool already watermarks outputs, do I still need to disclose deepfakes?
The provider's machine-readable marking under Article 50(2) and the deployer's deepfake disclosure under Article 50(4) are separate obligations on their face, and one does not state that it satisfies the other. If you deploy deepfake content, the 50(4) disclosure question is yours to assess — fact-specific, and one to confirm with qualified counsel.
See what may apply to your business
Answer seven quick questions for an automated, informational indication of which Article 50 obligations appear likely to apply — free, and not legal advice.