Does AI content have to be labelled?
In the EU, transparency rules apply to certain AI content and interactions. They are not the same for every case.
What is it?
In the EU, transparency rules apply to certain AI content and interactions: people must know they are talking to AI, deepfakes must be labelled and synthetic content should carry a machine-readable mark. The rules differ from case to case and apply in stages.
An everyday example
An online shop has to make clear that its chat is staffed by AI, not a person.
Why does it matter?
Labels help separate what is real from what is generated and help preserve trust in information.
A common misconception
That every sentence AI helped with has to be labelled, or that the absence of a label proves authenticity.
In more detail
What applies to talking with an AI?
In the relevant cases, users must be told that they are talking to an AI system.
What applies to deepfakes?
Certain synthetic or altered recordings carry disclosure obligations, with legal exceptions and details.
What is a machine-readable label?
Information that can help systems recognise that content was created or altered with AI.
Does every sentence AI helps write have to be labelled?
There is no simple rule for every use. The obligations depend on the purpose, the role, editorial control and the type of publication.
Remember: The labelling rule depends on the specific use.
Reliable sources
Last reviewed 1 November 2026. An explainer, not professional, legal or medical advice. For rules and figures that change, check the official version.