What was presented
Thirteen organisations and companies from Austria's media, advertising and communications industry have presented joint principles for labelling artificial intelligence in advertising. Supporters include the CCA Creativ Club Austria, the specialist associations of the Austrian Economic Chamber, the IAB, the MAV, the MCÖ, the OOHA, ORF-Enterprise and the VRM. According to the source ad-hoc-news.de, the rulebook is meant to create a uniform understanding of the transparency obligations under the European AI Regulation. These obligations have applied since 2 August 2026.
No obligation for every use of AI
According to the principles, the mere use of AI in production does not by itself trigger a labelling obligation. According to the source, blanket labelling of all AI-generated or AI-altered advertising is not planned, in order to avoid over-labelling. What matters is whether content qualifies as a deepfake.
Criteria for a deepfake
According to the European Commission, a deepfake exists when four criteria are met at the same time: a high resemblance to a real or plausibly existing reference, the false appearance of authenticity, the false appearance of truthfulness, and a disappointment of the audience's expectations. An explicit intent to deceive is not legally required.
According to the source, cases subject to labelling are expected to be relatively rare in advertising practice. An obligation arises, for example, when a real person is realistically portrayed as a testimonial. According to the source, purely artificial figures or synthetic voices that cannot be assigned to a real person generally do not fall under it.
Requirements for audio and images
If an advertisement must be labelled as a deepfake, the notice must appear at the latest at first contact and must be clear and directly related to the content. For audio formats, the principles provide either a spoken notice or a uniform sound signal; such an acoustic signal, a so-called earcon, is to be coordinated across the industry. For image and video material, a visual symbol should be used, for example the EU AI icon.
Assessment
The principles are a voluntary commitment by the industry and do not replace the interpretation of the AI Act by authorities or courts; the sources cite no official statement on this. For companies producing advertising with AI-generated material, it could make sense to check case by case, before publication, whether content meets the four criteria above. From the available sources it remains open how the case-by-case review will work in practice.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
