Retail giants including Amazon, H&M, Inditex, and Ikea are lobbying the European Union to exempt AI-generated advertising from the bloc’s upcoming deepfake transparency rules, arguing that product images and marketing materials pose no risk of deception and should not be burdened by the same labeling requirements as synthetic media designed to mislead. The push, led by European trade association Eurocommerce, directly challenges the EU AI Act’s definition of a “deepfake” and threatens to water down one of the law’s most visible consumer protections before it even takes effect.
Eurocommerce Argues Sofa Ads Are Not Deepfakes
In a letter to EU tech commissioner Henna Virkkunen, Eurocommerce Director General Christel Delberghe contends that an AI-generated image of a living room used to showcase a sofa should not fall under the EU’s definition of a deepfake. The association argues that requiring transparency labels on such content would affect a massive share of online advertising, ultimately diminishing the value of the rule for consumers who genuinely need protection from deceptive synthetic media. The EU Commission has not yet responded to the demand.
The EU AI Act, which takes effect on August 2, mandates clear labeling of AI-generated or AI-altered content that qualifies as a deepfake. The law defines deepfakes broadly, covering AI-generated images, audio, and video, with specific labeling tiers depending on whether the content is fully AI-generated or partially AI-modified. The regulation requires a base “AI” icon for any deepfake content, a “Fully AI-generated” label for content created entirely by AI without human creative input, and a “Partially AI-modified” label for pre-existing human-created content that has been altered using AI.
Retailers Already Depend Heavily on AI-Generated Content
The industry’s reliance on AI-generated marketing materials is already substantial and growing rapidly. Zalando reports that 90 percent of the marketing content on its platform is now AI-generated. Matthias Haase, VP of Content Solutions at Zalando, explains that generative AI has allowed the company to move from a planning mindset to a reactive one, cutting weeks of work down to just a few days, with a target of under 24 hours from spotting a trend to going live. H&M and Zara already use AI-generated clones of models for their online catalogs, replacing traditional photo shoots with synthetic imagery.
The scale of this adoption makes the EU’s transparency requirements particularly burdensome for the retail sector. If every AI-generated product image, model photo, or lifestyle shot required a label, the sheer volume of labeled content could overwhelm consumers and render the system meaningless. Eurocommerce’s argument hinges on the idea that a sofa image is fundamentally different from a deepfake video of a politician saying something they never said, and that treating them the same undermines the regulation’s purpose.
The Deepfake Label Is a Poor Fit for Commercial Imagery
The EU’s use of the term “deepfake” in this context is questionable. The word originated in the context of non-consensual pornography and is now primarily associated with fraud, identity theft, and other criminal activities. Applying the same label to an AI-generated product image of a sofa highlights just how blurry the current rules are. The regulation lumps together genuinely harmful synthetic media with benign commercial imagery, creating confusion about what the transparency requirement is actually meant to protect consumers from.
This conflation raises a fundamental question about the EU AI Act’s approach to AI-generated content. The law treats all AI-generated or AI-altered content as potentially deceptive by default, requiring labels regardless of context or intent. But a consumer looking at an AI-generated sofa image is unlikely to be deceived in the same way as someone watching a deepfake video of a political figure. The risk profile is entirely different, and the regulation fails to account for this distinction.
What the EU AI Act’s Deepfake Rules Actually Require
The EU AI Act’s transparency requirements for deepfakes are tiered based on the degree of AI involvement. For fully AI-generated content, the label must state “AI Generated” and apply to content created entirely by AI without human creative input or editorial control beyond prompting. For partially AI-modified content, the label applies when pre-existing human-created content has been modified using AI, such as replacing a face in a real photo with AI or furnishing an empty apartment using AI. A base “AI” icon is used when AI was involved in the creation of deepfake content but a custom text label or interactive second layer is also employed.
The regulation applies to images, audio, video, and published text on topics of public interest. This broad scope means that virtually any AI-generated marketing content could fall under the requirements, from product photos to model images to lifestyle shots. The retail industry’s argument is that this scope is too broad and that the regulation should focus on content that is actually intended to deceive, not on commercial imagery that consumers understand is promotional.
What This Means for the Future of AI Advertising Regulation
The outcome of this lobbying effort will have significant implications for how AI-generated content is regulated in the EU and potentially beyond. If Eurocommerce succeeds in securing an exemption for non-deceptive advertising, it could set a precedent for other industries to seek similar carve-outs. Conversely, if the EU Commission holds firm, retailers will face a significant compliance burden that could slow their adoption of AI-generated marketing content.
For businesses operating in the EU, the immediate takeaway is clear: the August 2 deadline is approaching, and the current rules apply to a broad range of AI-generated content. Companies using AI for marketing materials should prepare for compliance by implementing labeling systems that can distinguish between fully AI-generated, partially AI-modified, and human-created content. The debate over exemptions may continue, but the law as written is the current reality. Monitoring the EU Commission’s response to Eurocommerce’s letter will be essential for any organization that relies on AI-generated advertising in the European market.
