The EU AI Act and AI-Generated Product Images: Disclosure Explained

If you sell in the EU and use AI-generated product photos, the EU AI Act’s Article 50 sets a transparency duty: AI-generated or AI-manipulated image content should be detectable as such, and people should be able to know when what they are looking at is synthetic. The core obligations under Article 50 apply from 2 August 2026. This guide explains what that means in plain terms for an online fashion shop, and how a responsible service handles disclosure.

This is a regulatory explainer, not legal advice. The AI Act is a new and evolving regime, interpretive guidance is still being finalised, and your specific obligations depend on your role and setup. For decisions that carry legal weight, consult a qualified lawyer.

Short answer: do I have to disclose AI-generated product images?

In most cases where AI generates or meaningfully alters an image, yes, there should be a way for that to be known. Article 50 of the EU AI Act introduces transparency obligations for “certain AI systems,” and AI-generated or manipulated visual content falls within its scope. The duty splits across two roles:

• Providers of generative AI systems must mark their outputs in a machine-readable way so the content can be detected as artificially generated or manipulated.
• Deployers (the businesses using such systems) have disclosure duties in specific situations, most notably for deep fakes and certain public-interest content.

The practical takeaway for an online shop: treat AI-generated product imagery as something to be open about rather than something to hide. The Act’s stated purpose is to address deception and protect the integrity of the information people see, per the European Commission’s official Article 50 text.

What does EU AI Act Article 50 require?

Article 50 groups transparency obligations into a few buckets. The ones most relevant to ecommerce imagery are:

• Marking of synthetic output. Providers of generative AI systems must ensure outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. The marking has to be effective, interoperable, robust and reliable as far as technically feasible.
• Disclosure of deep fakes. Deployers who generate or manipulate image, audio or video content that constitutes a “deep fake” must disclose that the content has been artificially generated or manipulated.
• Interacting with an AI system. Where a person interacts directly with an AI system (for example a chatbot), they should be informed of that, unless it is obvious.

For a fashion brand, the headline is the marking-and-detectability principle: AI-generated visuals should carry the technical signals that let them be recognised as AI-made. The exact contours for “ordinary” product photography (as distinct from deep fakes of real, identifiable people) are still being clarified through Commission guidelines and a Code of Practice on AI-generated content. Several law firms tracking this, including Herbert Smith Freehills Kramer, note that the implementing detail is being worked out ahead of the deadline, which is one more reason to take a conservative, disclose-by-default stance.

When does the AI Act transparency rule take effect?

The Article 50 transparency obligations apply from 2 August 2026. The AI Act entered into force in 2024 and phases in over several years; the transparency provisions are one of the later milestones rather than something already binding.

There is a transitional nuance worth knowing. Under the EU’s “AI Omnibus” provisional agreement from May 2026, generative AI systems already placed on the market before 2 August 2026 were given additional time, until 2 December 2026, to meet the machine-readable marking requirement. That extension is about giving existing systems time to implement marking; it does not change the underlying direction of travel. The sensible planning assumption for a brand is that AI-content transparency is becoming the baseline expectation in the EU, not an optional extra.

How should an online shop label AI-generated photos?

There is no single mandated phrase, and labelling for product imagery is an area where guidance is still maturing. A practical approach for a small or mid-size fashion brand is to combine clear human-readable disclosure with respect for any machine-readable marking your tooling provides:

• Say it plainly somewhere accessible. A short, honest statement that product images are created with AI is clear and defensible. You do not need to plaster a watermark across every PDP; a clear note on the relevant pages or in your media/terms is a reasonable starting point.
• Keep machine-readable marks intact. If your generation tool embeds metadata or other markers that flag content as AI-generated, avoid stripping them out in post-processing or export.
• Be specific and non-deceptive. The aim is that a reasonable person is not misled into thinking they are looking at a conventional photograph of the exact physical item when it is an AI rendering. Describe what the image is.
• Document your process. Keeping a simple record of which images are AI-generated and how they are reviewed helps you answer questions from marketplaces, partners or regulators.

Marketplaces add their own rules on top of the law. Many platforms already publish image standards (for example requirements for a plain white product background), and some are introducing their own AI-content labelling. Disclosing openly tends to keep you onside with both the law and platform policy at once.

How Packshot Studio handles disclosure

Packshot Studio makes clean on-white packshots from the product images you already have, with no reshoot and no sample shipping. We are direct about how that works: the images we deliver are AI-generated and reviewed by a human before delivery. We do not present our output as human studio photography.

On fidelity, we aim for images that faithfully represent your real garment. We can’t promise every output is identical to a studio photograph, but we won’t deliver something that misrepresents your product, and a person checks the output before it reaches you. You can read more about how AI product photos are quality-checked and about AI product photography for fashion brands.

Because we are open about the method, brands can disclose clearly: you know the images are AI-generated and human-reviewed, so you can describe them honestly to your own customers and to the platforms you sell on. Packshot Studio is operated in the EU under GDPR by LuVi ApS (Denmark), and was founded by Ludvig Isaksen, founder of the Copenhagen label FINE CHAOS. How we handle the images you send us is set out in our privacy policy.

Where this leaves a fashion brand

The direction is straightforward: AI-generated visuals are increasingly expected to be marked and disclosed in the EU, with Article 50’s core obligations applying from 2 August 2026 (and a marking extension to 2 December 2026 for pre-existing systems). Working with a service that is transparent about AI generation and human review makes that disclosure simpler to honour. None of this replaces tailored legal advice; treat it as a starting map, then confirm the specifics for your business with a professional.

This guide was drafted with AI assistance and reviewed by a human before publishing.

Sources: European Commission — Article 50 (AI Act), Herbert Smith Freehills Kramer — Transparency obligations for AI-generated content.