The EU doesn't really know what a deepfake is, and that's becoming a problem for retail
Quick Answer
Eurocommerce, representing major retailers like Amazon and H&M, seeks exemptions for AI-generated ads from EU AI Act transparency rules, arguing that AI-generated images, such as a living room for sofa sales, shouldn't be classified as deepfakes.
Quick Take
With Zalando reporting that 90% of its marketing content is AI-generated, the EU's unclear definition of deepfakes poses challenges for the retail sector.
Key Points
- Eurocommerce advocates for AI-generated ads to be exempt from EU AI Act rules.
- Zalando claims 90% of its marketing content is AI-generated.
- The EU's lack of clarity on deepfakes complicates retail advertising strategies.
- Major retailers like IKEA are affected by these regulatory uncertainties.
- Misclassification of AI content could hinder innovation in marketing.
📖 Reader Mode
~2 min readEuropean trade association Eurocommerce is pushing to exempt AI-generated advertising from the EU AI Act's transparency requirements.
In a letter to EU tech commissioner Henna Virkkunen, which Reuters obtained, the association - whose members include Amazon, H&M, Inditex, and Ikea - is calling for an exemption for ad images that aren't meant to deceive. The EU law takes effect on August 2 and requires clear labeling of AI-generated or AI-altered content that qualifies as a "deepfake".
| Icon | When it is used | Examples |
|---|---|---|
| Base Icon (AI) | When AI was involved in the creation of deepfake content (image, audio, video) or published text, or when a custom text label or an interactive second layer is used | Deepfake video with the text label "Voices generated with," followed by the base icon |
| Fully AI-generated (AI Generated) | If the entire deepfake content (image, audio, video) or text was created entirely by AI, without any human creative input or editorial control (other than prompting) | Fully AI-generated deepfake videos featuring politicians or fictional events, AI-composed music or art, AI-generated news summaries |
| Partially AI-modified | When pre-existing, human-created content has been partially modified using AI, thereby becoming a deepfake or text on topics of public interest | A face in a real photo is replaced by AI with the face of a politician; authentic photos of an empty apartment are furnished using AI |
Director General Christel Delberghe argues that an AI-generated living room image used to showcase a sofa shouldn't fall under this definition. Requiring labels on that kind of content would affect a massive share of advertising and water down the transparency rule's value for consumers. The EU Commission hasn't responded to the demand yet.
Sofa images aren't deepfakes
Zalando says 90 percent of the marketing content on its platform is now AI-generated. "Generative AI […] allowed us to move from a 'planning' mindset to a 'reacting' one, cutting those weeks of work down to just a few days — and our target this year is under 24 hours from spotting a trend to going live," says Matthias Haase, VP of Content Solutions at Zalanda. H&M and Zara use AI-generated clones of models.
The EU's use of the term "deepfake" here is questionable at best. The word has its roots in non-consensual pornography and is mostly associated with fraud or other criminal activity. The fact that an AI-generated product image of a sofa falls under the same definition shows just how blurry the current rules really are.
— Originally published at the-decoder.com
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.
