Do you use AI to create or modify images, video, or audio in an advertising campaign? If so, you will soon be required to disclose in your advertisement that the content was generated by AI. The European Commission recently published draft guidelines confirming that the deepfake transparency obligation under the AI Act also applies to advertising. This is welcome clarification that advertising agencies and major brands have been awaiting for some time.

The AI Act requires deployers of AI systems to disclose that deepfake content has been artificially generated or manipulated (Article 50(4) AI Act). This is commonly referred to as the deepfake labelling requirement. A deepfake is AI-generated or AI-manipulated image, audio, or video content that closely resembles existing persons, objects, places, entities, or events and may falsely appear authentic or truthful to viewers. In principle, this obligation applies to all professional uses of AI, including advertising.

The labelling requirement will apply from 2 August 2026. Deepfake content that was deployed before that date does not need to be labelled retroactively.

The European Commission has now published draft guidelines that provide advertisers and advertising agencies with greater clarity on several important points:

• The concept of a deepfake is interpreted broadly. It does not only cover AI-generated content that imitates an existing person. Other AI-generated or AI-manipulated images, audio, or videos that appear realistic may also fall within the scope of the labelling requirement, including when used in advertisements.
• There is an exception for clearly creative, artistic, satirical, or fictional content. This exception may also apply to advertising, but the threshold is high: it must be immediately obvious to the average viewer, without any explanation, that the content is creative or fictional. A realistic advertisement featuring a recognisable person will almost certainly not qualify for this exception.
• The disclosure must be clear, prominent, and understandable. It must be noticeable and appropriate for the medium used. A disclaimer hidden in fine print or playback information is insufficient. For video content, the label should already be visible at the beginning of the video.

The guidelines are not yet final and do not currently have legal force. However, it is unlikely that they will change materially, and regulators and courts are expected to rely on them when interpreting and applying the AI Act.

One important point should not be overlooked: an AI label is not a free pass. A company may fully comply with the AI Act and still breach advertising laws. Existing rules on misleading advertising and unfair commercial practices continue to apply alongside, and in addition to, the AI Act.

Would you like to know what these new obligations mean for your advertising campaigns? We would be happy to discuss them with you.