You click on the page of a social network of your choice and a flood of images and videos surrounds you. In the middle of this riot of color, there is a sign: “Created with the help of AI.” We may soon see this little addition more often, as the EU is introducing new regulations for labeling AI-generated content. absatzwirtschaft spoke to Sebastian Deubelli, a specialist lawyer for copyright and media law and partner at SLD Intellectual Property Rechtsanwaltsgesellschaft, about the upcoming regulations, who they affect and whether companies should stay away from AI images in marketing.
Don’t: Rely on copyright for AI-generated images
According to Section 2 of the Copyright Act, only “personal intellectual creations” enjoy copyright protection. According to Sebastian Deubelli, human input is usually minimal in AI-generated images – the majority of the creative work is carried out by AI. However, a program or its provider cannot, in principle, obtain copyrights to generated images. “To be on the safe side, one should therefore assume that AI-generated images do not enjoy copyright protection,” says the expert.
First of all, this means that from a copyright perspective, images can simply be copied and used for one’s own purposes. However, in terms of marketing campaigns, this is not necessarily a deal-breaker: According to Deubelli, there is often probably no danger at all that competitors will simply copy the images from your own campaign.
Do: Adhere to obligations or better: mark them on your own responsibility
On August 1st of this year, the EU AI Act came into force. Among other things, this provides for a new labeling requirement for AI-generated images, which will take effect 24 months after the EU regulation comes into force. “However, the labeling requirement does not apply without restriction to AI-generated content and, in the area of images, refers to the creation and manipulation of content in the form of deepfakes,” explains Deubelli. Deepfakes are artificially created or manipulated content.
The responsibility for labeling lies primarily with the providers and operators of AI systems, while end users are not directly affected at first. SLD Intellectual Property nevertheless strongly advises dealing openly with content generated by AI and labeling it as such on your own responsibility.
Sebastian Deubelli advises on legal issues as a specialist lawyer for copyright and media law and partner at SLD Intellectual Property. (© Alexey Testov)
Don’t: Underestimate the challenges of detecting deepfakes and the like
In order to be able to cope with the abundance of material on the Internet, the AI Act stipulates that AI tools automatically generate labeled content. Meta has already recently implemented this. “However, the topic is still in its infancy and, especially in the area of fake news, we will probably have to assume that those who spread manipulated images will not point this out,” says Deubelli. “It is becoming increasingly difficult to distinguish these from real images, so in the end we will probably only have our gut feeling and a good dose of media literacy to identify deepfakes.”
The new regulation requires that content generated by AI systems be labeled in such a way that machines and people can recognize that it was artificially created or modified. Providers of AI solutions must ensure that their systems are reliable and user-friendly. They should also pay attention to costs and technical standards. For many normal users who do not fall directly under these rules, it would make sense, according to the specialist lawyer, if even a simple text indicated that the content came from an AI.
Providers or operators of an AI system who do not comply with the obligations under the AI Act primarily face fines. Sebastian Deubelli also points out: “In the case of AI-generated campaigns, the lack of a corresponding reference to the use of AI could violate the law against unfair competition. In this case, claims for injunctive relief and damages could be asserted, especially by competitors.”
Do: Use AI in marketing – but carefully
According to the expert, a completely legally secure use of AI is currently not possible. However, this does not mean that companies have to stay away from AI-generated images in marketing, says Deubelli: “The legal clarification will probably take years. Therefore, I am currently advising my clients not to wait until then, but to approach the topic proactively.”
If you consider the biggest legal risks, choose the right tools and approach the issue with a solid strategy, he sees great opportunities for companies at the process level but also in the context of content creation. “In my opinion, the risks can be reduced to an absolutely acceptable level.”
Laura Schenk
Laura Schenk (ls, born in 2002) has been a working student at absatzwirtschaft since August 2023 and is always keen to devote herself to new subject areas. She has a particular fondness for cubist painting and writing in all its forms. In 2023, she even dedicated a volume of short stories to her hometown of Leipzig. She is studying communication and media studies there and is involved in cross-media work for local newspapers and radio.
