The transparency obligations under Article 50 of the EU AI Act have applied since 2…

← Alle Beiträge
Künstliche Intelligenz · EN

The transparency obligations under Article 50 of the EU AI Act have applied since 2 August 2026.

This is a timely reason to examine a practical question: which AI-generated or manipulated content must actually be labelled?

For images, audio and video, the relevant obligation concerns deepfakes.

This may include:

1) photorealistic portraits resembling an existing or plausibly existing person;
2) artificial scenes resembling real places, entities or events;
3) materially altered photographs that appear to document a situation that did not occur;
4) realistic AI avatars; and
5) cloned voices that create the impression that a person made statements they did not make.

Ordinary editing, such as routine colour correction, is not treated in the same way. The key issue is whether AI-generated or manipulated content could falsely appear authentic or truthful.

Text is subject to a different test. AI-generated text published to inform the public about matters of public interest is, in principle, covered. However, the disclosure obligation does not apply where the content has undergone human review or editorial control and a natural or legal person assumes editorial responsibility for its publication.

Purely promotional text will typically fall outside this particular provision. Mixed promotional and substantive informational content may require closer assessment.

The attached European Commission Guidelines provide practical guidance on the implementation of the transparency obligations under Article 50 of the EU AI Act.

Zu diesem Beitrag gibt es auf LinkedIn ein Dokument als Anhang. Auf LinkedIn ansehen →

Auf LinkedIn diskutieren