AI Models in Advertising Rules: Ecommerce and Fashion

AI models in advertising rules come down to one principle in 2026: you may use an AI-generated model, but you must not let viewers believe a real person wore, tried or recommended the product when they did not. New York now requires ads to disclose synthetic performers, the EU requires professional deployers to label realistic AI content, and the FTC treats an AI model's praise as an endorsement that must be honest.

Checked October 2026 against official sources: the New York Senate bill S8420-A and the New York Governor's announcement, the FTC's Endorsement Guides, the EU AI Act Article 50, and the Meta, Etsy and Amazon pages linked below. This is general information, not legal advice; get advice for a specific campaign.

Ecommerce and fashion brands are the heaviest users of AI models: virtual try-on, AI lookbooks, and AI-generated "customers" holding products in UGC-style ads. This guide maps the rules by market and marketplace, then gives you copy-paste disclosure lines. For the wider legal picture of running AI personas, start with our AI influencer legal and ethics guide.

Who this guide is for

It is written for three groups: brands and store owners replacing or supplementing photo shoots with AI models, creators running AI personas that take paid brand deals, and agencies producing AI UGC ads for clients. If you only post AI content organically with no product attached, the platform labelling rules in our Instagram AI label guide matter more than advertising law.

The questions are the same for all three. Does the ad show a person who is not real? Does that person appear to recommend, wear or use the product? Does the image accurately show what the buyer gets? And where will the ad be seen?

AI models in advertising rules: how they stack, end to end

No single law covers AI models in ads. Several layers apply at once, and each asks a different question. Run every AI-model creative through them in order.

The first check is the biggest risk and the easiest to avoid. Training a model on a real model's photos, or swapping a real face onto an AI body, creates likeness and consent problems that a disclosure does not fix. Our guide to AI likeness rights covers releases.

Is there an AI-generated model ads law?

Yes, in New York, and it is the clearest one so far. New York amended General Business Law section 396-b through bill S8420-A. The Governor's office announced on June 9, 2026 that the law, signed in December 2025, was now in effect.

The bill text requires any person engaged in selling property or services who, for a commercial purpose, produces or creates an advertisement to conspicuously disclose in that ad that a synthetic performer is in it, where the person has actual knowledge. A violation carries a civil penalty of $1,000 for the first violation and $5,000 for each later one.

The definition fits AI fashion models closely. A synthetic performer is a digitally created asset, created, reproduced or modified with generative AI or a software algorithm, intended to create the impression of a human performer who is not recognizable as any identifiable real performer. So an original AI model in a dress ad is squarely covered. The bill excludes audio-only ads, AI used solely to translate a human performer, and ads for expressive works such as films and games where the synthetic performer use matches the work. Media outlets that carry the ad are generally not liable unless they had written notice.

What AI fashion model regulations apply in the EU?

The EU has no fashion-specific rule, but the AI Act's transparency article reaches most realistic AI-model ads. Article 50(4) requires deployers to disclose deepfakes, and the definition in Article 3(60) covers AI content that resembles existing persons, objects, places or events and would falsely appear to be authentic. The European Commission's Article 50 FAQ says the person or thing simulated must exist or could plausibly exist, which captures a photorealistic model who never existed but looks like any shopper.

A business running ads is acting professionally, so it is a deployer, and the duty has applied since August 2, 2026. The disclosure must be clear and distinguishable and reach viewers at first exposure, and the Commission says machine-readable watermarks alone do not satisfy it. Our guide to EU AI Act deepfake labeling walks through the tests and the dates. EU consumer protection law on misleading advertising applies alongside it.

AI models as endorsers: the FTC view

In the US, the bigger risk for most brands is not a missing AI label but a misleading endorsement. The FTC's Endorsement Guides, 16 CFR Part 255, define an endorsement as a message consumers are likely to believe reflects the opinions or experiences of a party other than the advertiser, and say the endorser could be or appear to be an individual. The FTC's 2023 announcement said the updated definition clarifies that it includes virtual influencers.

Two consequences follow. First, an AI model saying "this serum changed my skin" is an endorsement, and since an AI has no skin, the claim describes an experience nobody had. Second, the FTC's 2024 final rule on fake reviews and testimonials addresses testimonials that misrepresent that they are by someone who does not exist, such as AI-generated fake reviews, or by someone with no actual experience of the product. An AI "customer" presented as a real buyer sits right in that zone.

The FTC's Disclosures 101 for Social Media Influencers adds practical rules that apply equally to AI creators taking brand deals: put the disclosure where people will see it, superimpose it on image-based posts, put it in the video itself rather than only the description, and do not bury it in a group of hashtags.

AI model disclosure for ecommerce: marketplace by marketplace

Platform policies change faster than laws, so treat this table as a snapshot of the official pages as of October 2026, then confirm on the page itself before a big campaign.

Meta. Meta's help page on AI info for ads says it automatically labels ad content created or edited with some of its generative AI tools, such as Background Generation, Image Generation or Add Animation, and with third-party generative AI tools; the page notes the feature may not be available in every region. Viewers find the label by opening the three-dot menu and choosing About this ad, which Meta's ads transparency announcement (updated June 2026) describes as the central place for AI info on ads. Because the label depends on Meta's tools or detectable signals, an AI image made elsewhere without provenance data may carry no label at all. For social issue, election or political ads, advertisers must disclose photorealistic AI content, including a realistic-looking person who does not exist.

Etsy. Etsy is the strictest on images. Its Creativity Standards say listing images must be of the item itself, not a rendering or stock photo, and that seller-prompted AI creations must be disclosed in the listing description. Its Seller Policy asks for your own photographs or video, not stock photos or artistic renderings, and requires disclosure in relevant listings when an item is created with AI. An AI model wearing a garment that was never photographed is hard to square with those rules.

Amazon. We did not find a public Amazon rule requiring AI disclosure on listing images. Amazon's product image requirements, published in Seller Central for logged-in sellers, require images that accurately represent the product, and category style guides add their own rules. On the ads side, Amazon Ads offers AI image generation in its ad console, and its ad specs say that opting in to AI enhancement can add AI-generated content to your creatives. Check your category style guide before using AI models on apparel listings.

Copy-paste disclosure lines

Short, plain lines work best. Put one on the image or video itself and repeat it in the first line of the caption or ad text. Adjust the wording to the facts; do not claim more or less AI use than happened.

• AI model, real product: "AI-generated model. Product shown is real."
• Fully AI image: "This image was created with AI."
• Virtual try-on: "Digitally generated try-on. Fit may vary; see size guide."
• AI persona in a paid post: "Ad: paid partnership with [Brand]. I'm a virtual creator; images are AI-generated."
• AI voice-over: say "This ad uses an AI-generated voice" at the start, and add it on screen.
• Etsy listing description: "Listing images include an AI-generated model; the item itself is photographed in images 2 to 5."

Rolling this out across a catalogue

If you are building AI models or a virtual creator to sell products, our AI Influencers program covers persona design, consistent characters and brand deals, with disclosure built into the workflow from the first post. For the ad production side, our AI UGC ads guide covers formats and tools.

Common mistakes

• Training on a real model's portfolio. A "new" AI face built from one person's photos can still be recognisable. That moves you out of synthetic performer territory and into likeness law; see our deepfake laws by country table.
• Relying on the platform label. Meta's automatic label depends on its tools or detectable signals. If your AI image came from elsewhere and carries no metadata, no label appears.
• Disclosing AI but not the ad. "AI-generated" does not tell anyone a post is paid. Sponsored AI content needs both.
• Improving the product. AI that smooths a fabric, changes a colour or tightens a fit can misrepresent what buyers get, even with a label.
• Forgetting reposts. Caption disclosures vanish when an ad is screenshotted; on-image labels survive.

Pre-launch checklist

AI models in advertising: FAQ