THE GUIDE TO SHOP IMAGES
AI in the picture? Learn when a disclosure may be needed, how to make it visible and what you should check.
Three questions about your imageUpdated 5 October 2026 · Based on the EU Regulation and Commission guidance
YOUR IMAGE CHECK
Three questions give you an initial indication. No sign-up. Answer for the specific image and how you use it.
Your answers stay private. They are never saved or shared.
Read the rules in detailYOUR INITIAL GUIDANCE
Guidance based on your answers, not a binding legal assessment or confirmation of compliance.
02 /
The AI Act covers, among others, businesses in the EU. It can also cover providers and deployers outside the EU when their AI output is used in the EU. Serving EU customers is therefore a reason to check the scope, not a blanket labelling duty for every shop.
Stores outside the EU should also check whether the rules apply when their AI content is used in the EU.
EU · Art. 2 ↗For images, the visible disclosure duty in Article 50(4) concerns deepfakes: AI-generated or manipulated depictions resembling real people, products, places or events and falsely appearing authentic. Something that could plausibly exist may also qualify. Context and audience matter.
Example: An invented, photorealistic product scene may be relevant. A clearly recognisable illustration needs a different assessment.
EU · FAQ ↗Where disclosure is required, the AI generation or manipulation must be clear and accessible by the first exposure. A clear, visible label can help. Hover or tap details add information; the required disclosure must not be hidden behind that interaction.
An icon alone is no guarantee. Check clarity, contrast, mobile display and keyboard or screen-reader access.
EU · Art. 50(4–5) ↗For evidently artistic, creative, satirical or fictional works, disclosure may be adapted so that it does not hamper their display or enjoyment. This is not a blanket exemption. Ordinary image corrections are not automatically deepfakes either. The standard-editing exception in paragraph 2 concerns machine-readable marking by AI providers.
EU · Art. 50(2, 4) ↗When downloading, sharing or using the image in an ad, check the disclosure in that setting again. aiconit does not alter the original file. An AI label does not make misleading product depictions permissible.
Other duties concern AI chatbots, machine-readable marking by generative AI providers, emotion recognition and biometric categorisation, and certain AI texts on matters of public interest. This guide focuses on shop images. The transparency duties apply from 2 August 2026.
EU · Art. 50 ↗03 /
The product is real, but its surroundings were generated with AI. If you know that is how the image was made, mark it as containing AI content and choose your label.
An AI-generated person models a real product. Confirm the AI content based on the image’s known origin. aiconit displays your chosen label and leaves the original file unchanged.
You know this is a real photograph with no generative edits. Record that decision without an AI label. If the origin is unclear, leave the image unreviewed until you can confirm it.
With your permission, we use Google Analytics to measure visits and submitted launch signups, and Google Ads to measure advertising results. No personalised ads. Email addresses and form contents stay with us.
Privacy details