The EU doesn't really know what a deepfake is, and that's becoming a problem for retail
The EU AI Act includes provisions requiring that AI-generated or manipulated content be labeled as such, a measure primarily aimed at combating disinformation and synthetic media that misrepresents real people or events. But the regulation's definition of what qualifies as a "deepfake" is broad enough that it may also sweep in routine commercial imagery - product photos, lifestyle scenes, and promotional visuals that are generated entirely by AI rather than captured by a camera.
Eurocommerce, the Brussels-based trade association that counts Amazon, H&M, and IKEA among its members, is now formally asking regulators to carve out an exemption for this kind of content. The group's argument is straightforward: a digitally rendered image of a sofa sitting in an AI-generated living room is not, in any meaningful sense, the same thing as a fabricated video of a politician saying something they never said. Requiring retailers to label every such image as synthetic could be disruptive and, in their view, unnecessary given that no real person is being misrepresented.
The scale of the issue is illustrated by Zalando's disclosure that roughly 90 percent of the marketing content on its platform is already AI-generated. That figure reflects a broader industry shift - generative image tools have become deeply embedded in e-commerce workflows, where the speed and cost advantages of synthetic visuals over traditional product photography are difficult to ignore. If the Act's transparency obligations apply in full to this content, retailers would face a significant compliance burden and potentially confusing labeling requirements for consumers browsing ordinary product listings.
The underlying problem is a definitional one. The EU AI Act was drafted with harmful synthetic media in mind, but the language used does not cleanly distinguish between a fabricated political video and an AI-rendered furniture ad. How regulators interpret and clarify that language - whether through formal guidance, delegated acts, or amendments - will have a direct bearing on how generative AI tools are used in commercial contexts across Europe. The retail industry's intervention is an early sign that the practical consequences of vague AI legislation are arriving faster than the rule-making process anticipated.
