How we mark AI images

Last updated: 21 lipca 2026

This English text is a convenience translation. In case of any discrepancy, the Polish-language version is legally binding.

Every image generated in Renvido is automatically marked as AI-generated content in a machine-readable way, in line with the marking requirement of Article 50 of the EU AI Act.

How it works

  • We add an XMP packet to each file's metadata (JPEG/PNG) with the IPTC standard field Iptc4xmpExt:DigitalSourceType set to trainedAlgorithmicMedia (an image fully generated by an AI model) or compositeWithTrainedAlgorithmicMedia (a post combining an AI image with layout and text).
  • This is a standard read by Google, Meta, LinkedIn and Adobe, among others — a real, recognized marker of AI content.
  • The marker is invisible: we add no visible watermark. Your files stay clean and ready to use — the marking lives only in the metadata.
  • We embed it losslessly, at the file-container level — pixels are not re-compressed, so quality (up to 4K) is preserved.

How to verify the marking

You can check the marker with any tool that reads image metadata, e.g.:

  • on the command line: exiftool -DigitalSourceType file.jpg;
  • in an XMP/EXIF metadata viewer (e.g. "File info" in Adobe Photoshop, or free online metadata viewers);
  • in content-ecosystem tools (e.g. C2PA-compatible verifiers that read provenance fields).

Scope and responsibility

Renvido marks the generation output in a machine-readable way. The duty to inform your own customers or audience that a piece of content is AI-generated — where required by law or by a platform's policy in a given context — remains with the seller. Our marking helps you meet that duty but does not replace it.

Learn more about data processing in our Privacy Policy, and about using the service in our Terms.