From August 2026, the transparency obligations of the European AI Regulation will apply. What does this mean for creators who use AI images, AI voices or avatars?
What does the AI ACT regulate?
The European AI regulation is gradually entering into force. Most regulations, including transparency obligations, apply from August 2, 2026. Anyone who creates or distributes so-called deepfakes, i.e. realistic image, sound or video content that real people, places or events imitate, must reveal that the content is artificially generated or is manipulated.
What does this mean for creator?
For creators, it is particularly relevant whether content gives the viewers the impression of being real. An obviously artistically alienated picture is to be evaluated differently than a realistic video in which a real person says things he never said. Irrespective of the AI Act, the prohibition of misleading applies in advertising: Anyone who advertises cosmetics or fitness products with AI-embellished results is deceiving consumers.
- Mark realistic AI representations of real people
- Make AI Voices and Avatars Transparent
- Use platform labels for AI content
- Show no AI embellished results in advertising
Who owns AI content?
Only personal intellectual creations of people are protected by copyright. Purely AI-generated images are therefore regularly not protected. However, protection can consist of human selection, editing and composition.
Practical examples
AI Avatar for Cooperations
Situation: A creator has a digital twin that speaks promotional clips.
Classification: The avatar looks realistic. Viewers could assume the creator is speaking.
Recommendation: Mark avatar clearly and regulate the license with the agency in writing.
AI thumbnail
Situation: A YouTuber creates his thumbnails with an image generator.
Classification: Stylized thumbnails are not required to label deepfake, but misleading the video content is to be avoided.
Recommendation: Don’t make thumbnails misleading.
Deepfake with a stranger’s face
Situation: A creator wants to use the face of a celebrity in a sketch via AI.
Classification: In addition to the labeling, the personality rights of the celebrity are affected.
Recommendation: Without consent only in the narrow context of satire, if in doubt.
Conclusion
AI opens up new opportunities for creators. Anyone who works transparently and respects personal rights is also on the safe side after August 2026.
Do you have questions on this topic? More on this on our page AI influencer and avatars or directly via our Initial assessment.
Frequently asked questions
Do I need to label AI-generated content?
As of August 2026, the AI ACT requires the disclosure of Deepfakes and certain AI-generated texts on public interest issues. Platforms like Instagram and TikTok also require their own AI labels.
Who owns the rights to AI-generated images?
Purely AI-generated content usually does not enjoy copyright protection because there is a lack of human creation. Protection can arise when people decisively design or process the result.
Can I train an AI with a real person’s voice or face?
Only with consent. Voice and face are protected by personal rights; There is a risk of injunctive relief and damages without consent.
Further information: AI influencer and avatars · Copyright