The European Union finds itself in a difficult position: the AI Act imposes transparency requirements on synthetic content, but there is still no agreed-upon, operational definition of “deepfake.” This regulatory gap directly affects the retail sector, where the use of AI-generated images and videos is now standard practice. Zalando, for example, reports that 90% of the marketing content on its platform is already AI-generated.
Eurocommerce—the trade association representing giants like Amazon, H&M, and IKEA—has formally requested an exemption for AI-generated advertisements from the transparency rules laid out in the AI Act. The argument is simple: an image of a virtual living room used to sell a sofa is not a deepfake. However, the distinction between synthetic content
The context: when regulations fail to keep pace with technology
The European AI Act came into effect in 2024. Its provisions on synthetic content transparency are being applied progressively. However, the regulatory text has a significant loophole: it does not provide an operational and unambiguous definition of “deepfake” applicable in a commercial context.
This ambiguity is not a technical detail. It is a structural problem. In fact, the distinction between manipulative content—intended to deceive—and AI-generated images used purely for aesthetic and promotional purposes is not yet clearly delineated by regulations.
The issue has emerged strongly in recent weeks. The Decoder has documented Eurocommerce's position., the association representing Amazon, H&M, IKEA, and hundreds of other European retailers. The request is explicit: exempt AI-generated advertisements from the transparency obligations stipulated by the AI Act.
The numbers that matter: retail is already past the tipping point
The most significant figure comes from Zalando. 90% of the marketing content on the platform is already AI-generated. This is not a marginal experiment. It is the standard operating procedure for one of Europe’s leading digital retailers.
This number reframes the debate. We're not talking about an emergent phenomenon to regulate in advance. We're talking about an already established reality, which regulation struggles to frame retroactively.
Furthermore, the trend is set to accelerate. According to Gartner, by 2027, more than 30% of marketing messages from large organizations will be generated synthetically. In the retail sector, the percentage is already higher. As a result, pressure on European lawmakers to clarify regulatory boundaries is set to increase.
Among other things, the problem isn't limited to major players. Italian SMEs that use generative AI tools to produce product images, promotional videos, or social media content also find themselves technically in a gray area.
Eurocommerce's position: a defensible argument?
Eurocommerce maintains that an AI-generated image of a living room—used to contextualize a sofa in e-commerce—is not a deepfake. The argument has its logic. A deepfake, in its original meaning, is content created to make something false appear real, often with malicious or defamatory intent.
A photorealistic rendering of a domestic environment, however, does not deceive anyone about the nature of the product being sold. Therefore, applying the same disclosure obligations to both categories seems disproportionate.
However, the issue is not that simple. On the contrary, the line between synthetic content
Related articles
Discover other articles that explore similar topics in depth, selected to give you a more complete and stimulating view. Each piece of content is carefully chosen to enrich your experience.
