Labelling AI images in your online store under the EU AI Act

Key facts

• The label must be visible no later than first exposure, for every person and every place the image is shown (Article 50(5); para. 143). In a store that is often the product card, not the product page.
• The “AI” belongs on the image and must be visible without a click or hover (Code of Practice, Section 2, Sub-measure 1.2.1 a).
• A notice that only appears in terms, an FAQ or an information page is easy to overlook and does not suffice (para. 142).
• Google Shopping, Meta ads, marketplaces and newsletters are your responsibility. An overlay in your store does not reach them.
• For supplier and agency images where you took no decision on the use of AI, you are generally not the deployer. The guidelines still encourage keeping existing labels (paras. 14, 16, 17).
• A label does not make a misleading image lawful and does not replace the consent of people shown (recital 137; paras. 127, 129).

Where the label has to go

The yardstick is Article 50(5) of the EU AI Act: the information must be provided at the latest at the time of first exposure. The guidelines mean not only the first person who sees an image but every later one too. The duty applies to each output at any moment someone is reasonably likely to see it (para. 143). In an online store, that is rarely the product page first. More often it is a product card in a collection, a search result, a recommendation or a banner on the home page.

That leads to the overview below. The sizes are our recommendation. Neither the AI Act nor the Code of Practice sets pixel values; the Code asks for a label that stays clear and distinguishable at any size (Measure 1.1 c).

Thumbnails and small previews

There is no exception for scaled-down images. The guidelines do exclude icon-scale graphics from the marking duty under paragraph 2 (para. 68), but that concerns tiny graphics generated by AI, not a small rendering of a deep fake. On thumbnails, the “AI” without text is enough as long as it stays recognisable. For colour swatches and very small previews, the large image on the same page carries the label.

Hover images and variants

Many themes swap in a second image, often a lifestyle or model shot, when you hover over a product card. The label has to follow the image that is actually visible. If only the second image is AI-generated, it needs its own label as soon as it appears. The same applies to variant switches and sliders: the label belongs to the image, not to a position in the gallery.

Checkout, order confirmation and emails

Many e-commerce platforms do not allow overlays on product images in checkout or in order emails. Anyone who reaches checkout has almost always seen the image with its label in the store before. In our view, that covers first exposure for that person. A residual risk remains for direct links into checkout, such as “buy again” buttons in emails. Burning the “AI” into the image file reduces it (see AI label design).

What does not count

• A notice image in the gallery. A first gallery slide saying “Some images are AI-generated” labels none of the AI images. It does not appear on product cards, in search or in zoom, and visitors can skip it. The duty, however, applies to every place the image is shown (para. 143).
• A line in the product description. Descriptions and bullet points sit beside or below the image, often below the fold, and are missing from cards altogether. The information has to stand apart from other information and must not be easy to overlook (para. 142). For an image, that means on the image.
• Terms, FAQ, legal notice or an information page. The guidelines name terms of use, manuals and hidden menus as unsuitable (para. 142). An explanatory page can complement the label.
• A blanket “we use AI”. It does not say which image is affected. For chatbots, the guidelines call such general statements insufficient (para. 38), and the reasoning carries over to images.
• Metadata in the file. Nobody perceives machine-readable marking without tools (para. 117).
• A label that only appears on hover. The “AI” has to be visible without any interaction (Code, Sub-measure 1.2.1 a). Only additional detail may sit behind a click.

Channels outside your store

An overlay only works inside your own store. Wherever you reuse a deep fake yourself, you label it there too. Deployers in multi-step distribution chains should take proportionate steps so that their label reaches the audience, for example through contractual terms with distribution partners and interface settings (para. 12).

• Google Shopping and Merchant Center: Google requires the IPTC value “trainedAlgorithmicMedia” in AI image files. Since July 2026 it expressly allows your own AI labels in the image but advises against placing them in the outermost corners, which may be cropped. The general image specification still lists overlays as a reason for disapproval. Details under Google Merchant Center.
• Meta (Facebook, Instagram): Ads only get an automatic AI notice when Meta's own AI tools were used. Images from other tools are yours to label, for example with a version that has the label burned in.
• TikTok: Under its ad policy, AI-generated ad content needs the AIGC label or a clear disclaimer of your own. TikTok's exemptions for edits do not replace the Article 50 assessment.
• Amazon: Images with photorealistic, fully AI-generated people should carry the keyword “contains-synthetic-performer” in their metadata.
• Newsletters, marketplaces and your own social posts: Where no overlay is possible, use a version with the label burned in.

Within a very large online platform, deployers may rely on the platform's labelling tools if they allow a clear and distinguishable disclosure, but only within that platform (para. 126). Machine-readable entries are explained in AI metadata.

Supplier images, stock photos and agency work

Whether you are the deployer for someone else's image depends on who decided to use AI.

Three points always apply. If you edit someone else's image with AI, you are the deployer for that edit. Never remove an existing label or provenance data; the guidelines encourage keeping them (paras. 16 and 98). And consumer protection law does not ask who created an image: you are responsible for your advertising, borrowed images included. When in doubt, ask your supplier whether AI was used.

Documentation: process, record, corrections

Signatories of the Code of Practice set up internal processes and documentation describing how they label, in proportion to their size and resources (Commitment 2, Measure 2.1). Non-signatories are expected to show by other means how they meet the obligation, for example through a gap analysis against the Code, and may face more detailed requests for information (para. 148). Three building blocks help either way:

1. A short procedure. Who decides on labelling? Which AI tools does the team use? Who checks new images before publication? The Code suggests a general description and real examples. Those who create AI images regularly should also check that labels display correctly (Measure 2.1).
2. A record per image. Which image, when and with which tool it was made or changed, the decision with reasons and source, label on or off, and who decided. It also helps with older images: if you can show an image was created before 2 August 2026, it does not need a label retroactively (para. 154).
3. A correction route. Review reports of missing or wrong labels and fix them without undue delay (Code, Measure 2.3). A reporting channel for customers is encouraged by the Code, not required.

Staff and contractors involved in labelling should know about the obligation (Code, Measure 2.2). More in Code of Practice.

A label fixes nothing else

Labelling meets the transparency duty under Article 50. It does not make an image lawful. Recital 137 of the AI Act states that compliance with the transparency obligations should not be read as meaning that the use of the content is lawful. The guidelines name misleading advertising as an example of deep fakes that remain unlawful despite a label (para. 129).

If an AI image shows a product in a colour, size, quality or effect it does not have, that can be misleading under consumer protection law, with or without a label. The label is no substitute for an image that shows the product accurately.

Real people need to consent

If you use AI to change an image of a real person, such as a model's face, figure or skin, data protection and personality rights apply (paras. 127, 129). As a rule you need that person's consent. The Code states that labelling does not exempt anyone from other laws, expressly including consent of the people depicted (Section 2, Commitment 1). Check your model contracts before editing shoots with AI.

Typical cases

• AI model wearing the product
• Real product in front of an AI background
• Background removal with AI
• Colour variant generated with AI

All cases are collected under Cases. For Shopify there is an app that adds the label across all store surfaces and records each decision.

Not legal advice. This content explains the law based on primary sources; it does not replace an assessment of your individual case.