When to disclose that a fashion image was AI-generated

Short answer: a fashion image generated (or substantially modified) with AI should be disclosed whenever it could lead the audience to believe it depicts real people, places or products that aren't real. In childrenswear, the transparency threshold should sit higher than the legal minimum.

Table of contents

• Why disclosure matters in fashion
• The regulatory landscape
• When disclosure is mandatory
• When it is strongly recommended
• When it is not required
• The childrenswear-specific case
• How to word the disclosure
• Common mistakes
• The AISEOSA — Kids AI Fashion Studio method
• Operational checklist
• FAQ
• Conclusion

Why disclosure matters in fashion

Trust is an economic asset. The moment a consumer discovers that an image they thought real was AI-generated — and no one said so — the reputational damage almost always exceeds the production benefit. Disclosure isn't a cost: it's a form of positioning.

The regulatory landscape

• EU AI Act (Reg. 2024/1689): introduces a transparency obligation for synthetic content depicting people, places or events in ways that could be mistaken for real (deep fakes). Full application phases in between 2025 and 2027.
• UK ASA and FR ARPP: have published guidance requiring clear labelling when AI significantly alters the depicted reality.
• National consumer laws (e.g. Italy's Codice del Consumo): forbid misleading commercial practices. An AI image presented as a "product photograph" that doesn't match the real garment can fall inside that scope.
• CNIL (FR) and equivalent data protection authorities: add rules when the image concerns identifiable minors.

Practical rule: when the image could mislead, disclose. When it involves minors, disclose even before the law demands it.

When disclosure is mandatory

• Images depicting identifiable real people who never wore that garment or weren't in the depicted location.
• Synthetic models presented as real people (fictional testimonials with name, biography, socials).
• Campaigns telling an event (show, backstage, trip) that never happened.
• Anything falling under the AI Act's definition of deep fake.

When it is strongly recommended

• Anonymous synthetic models in editorial campaigns: even without a formal obligation, disclosure protects the brand from accusations of narrative greenwashing.
• AI locations: a villa, beach or city that doesn't exist, presented as real.
• Substantial modifications of original photos (face swap, age change, body reshape).
• Childrenswear e-commerce where the garment is real but the model is synthetic: disclosure reassures the parent-buyer.

When it is not required

• Standard post-production (colour, light, background clean-up) equivalent to analogue retouching.
• Product-only images with no model, where AI is used only to remove background or produce a neutral still-life coherent with the real garment.
• Internal use (moodboards, decks, research) not distributed to the public.

The childrenswear-specific case

In childrenswear, disclosure isn't optional even when the law doesn't explicitly demand it. The primary audience — parents — reacts badly to surprises. Operational guidance:

• Always disclose the use of AI models, both purely synthetic and digitised "human kids models".
• Distinguish product images (real garment) from narrative images (scenes, trips, contexts).
• Avoid any image that could suggest an identifiable real child who didn't actually participate.

How to word the disclosure

Useful formulas, tunable by channel:

• Discreet campaign caption: "Image created with AI support. Real garment. Digital model."
• E-commerce site: "Digital model. Real product, photographed and digitally reconstructed for colour fidelity."
• Social: explicit hashtag or platform-native tag (Meta, TikTok) when available, e.g. #AIassisted.
• Press kit: a dedicated paragraph explaining method, consent and limits.

Language rule: "AI support" or "AI-assisted production" avoids both understatement and sensationalism.

Common mistakes

• Strategic silence: hoping no one notices. Statistically doesn't work.
• Hidden disclosure buried in site legals, invisible in the campaign itself.
• Ambiguous label ("digitally enhanced") that doesn't clarify generative AI use.
• Reverse overclaim: labelling as AI even where it isn't, to look innovative.

The AISEOSA method

Every AISEOSA production ships to the client with:

1. An image classification (pure product, synthetic model, digitised model, scene, retouch).
2. A per-channel disclosure recommendation (campaign, e-commerce, social, PR).
3. A ready-to-use language kit in IT and EN.
4. A methodology note the brand can publish on its own site to reassure parents.

Operational checklist

• Does the image depict real people or events? If yes, disclosure is mandatory.
• Is the model synthetic or digitised? If yes, disclosure is recommended.
• Does the depicted product faithfully match the real garment?
• Is the wording understandable to a non-technical parent?
• Is the disclosure visible in the channel where the image appears?
• Is there a public methodology note supporting it?

FAQ

Does disclosure hurt sales?

Not in premium brands. On the contrary, transparent AI use is perceived as a sign of brand maturity.

Is "AI generated" in small print enough?

No. Disclosure must be legible, clear and close to the image, not hidden in the footer.

What if I only use AI to remove the background?

No disclosure needed: it's equivalent to traditional photo retouching.

Should I disclose AI in an internal B2B catalogue?

Not mandatory, but good practice — consistency between external and internal communication builds trust with retail partners.

Conclusion

Disclosing AI use in a fashion image is not an admission — it is a positioning move. In childrenswear, it raises trust with the most demanding audience (parents) and reduces reputational risk. The AISEOSA criterion is simple: if an image could be mistaken for a reality that doesn't exist, disclose it; if it depicts a minor, disclose regardless.

Related reading

• How to choose an AI studio specialised in childrenswear
• Product Preservation in AI Fashion: How to Protect Garment Details
• AI in Childrenswear: A Practical Guide for Fashion Brands

Sources

• Regulation (EU) 2024/1689 — AI Act, art. 50 and following.
• Italian Codice del Consumo, Legislative Decree 206/2005, art. 20–23.
• ASA (UK), guidance on AI-generated advertising, 2024.
• ARPP (FR), "Intelligence artificielle et communication commerciale" recommendation, 2024.
• Data protection authorities' provisions on minors' images online.