Guides · Article 50 · EU AI Act
Labeling AI-generated image outputs
Text-to-image tools now sit inside design apps, ad platforms, and everyday workflows. Two EU AI Act paragraphs can bear on their outputs: Article 50(2), the provider-side machine-readable marking of synthetic images, and Article 50(4), the deployer-side deepfake disclosure where an image depicts real people convincingly. This guide explains both from the regulation text. It is informational only, not legal advice, and does not name any generator as in or out of scope.
Provider marking — Article 50(2)
Article 50(2) asks providers of AI systems generating synthetic audio, image, video or text to ensure outputs are marked in a machine-readable format and detectable as artificially generated or manipulated, with solutions effective, interoperable, robust and reliable as far as technically feasible, taking into account the specificities and limitations of the content type. For images, that points toward marking that travels with the file — the paragraph does not name one technology. This duty is written on the provider of the generator, not on every user who downloads an image. Whether a given tool falls inside 50(2) is fact-specific.
Deployer deepfake disclosure — Article 50(4)
Where a generated image depicts real people in a way that constitutes a deep fake, Article 50(4) is written on the deployer of the system: they must disclose that the content has been artificially generated or manipulated. So a synthetic photo of an identifiable person raises a different, deployer-side question from a generic illustration. Where the content forms part of an evidently artistic, creative, satirical, fictional or analogous work, the paragraph limits the obligation to disclosing the existence of generated content in a manner that does not hamper enjoyment of the work. Whether your image is a 'deep fake' and whether a carve-out applies are fact-specific.
Two roles, two questions
The practical takeaway is that image outputs can implicate a provider-side marking question (50(2)) and, separately, a deployer-side deepfake question (50(4)) — and the same organisation can hold different roles for different systems. Mapping which applies, if either, is the useful first step. Our free scope check gives an informational read; confirm the analysis with qualified counsel.
Common questions
Do I have to label images I make with an AI generator?
The machine-readable marking duty in Article 50(2) is written on the provider of the generator; the deepfake disclosure in 50(4) is written on the deployer where an image depicts real people convincingly. Which applies to your case, if either, is fact-specific — verify 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.