In 2026, AI virtual models are no longer an experiment. They are part of how fashion brands shoot, launch and sell. Zalando, H&M, Mango and Zara have all brought AI-generated or AI-enhanced models into their e-commerce and campaign imagery, and for most brands the question is no longer "should we?" but "how do we do it well?".
Here is what the big players are doing, what works, what has gone wrong, and how a smaller brand can use virtual models without losing its identity.
1. What an AI virtual model is
An AI virtual model is a digital person, created or trained with generative AI, who can wear a collection in any pose, light or location. There are two main approaches. The first is a fully synthetic model: a face and a body that do not exist, designed around the brand. The second is a digital twin: the AI version of a real model, built from a set of reference photos and used with that person's consent and payment.
In both cases the logic is the same: a consistent face, a consistent body, and the freedom to produce new images without booking a new shoot every time.
2. How the big fashion brands use them
Zalando has been one of the most open about it. By the end of 2024 around 70% of the images in its editorial campaigns were made or enhanced with AI. In May 2025 its vice president of content, Matthias Haase, told Reuters that AI lets the company be "reactive" to trends: images in a few days instead of a six-to-eight-week shoot cycle, with costs down by up to 90%.
In March 2025 H&M announced digital twins of 30 real models. Each model keeps the rights to her own twin and is paid when it is used; the first images appeared in the summer of 2025.
Mango launched an AI-generated campaign for its Teen line in July 2024 and, in June 2025, began using AI images on product pages, where speed and volume matter most. At the end of 2025 Zara started using AI to re-dress existing photos of its models in new garments, with their agreement and compensation, alongside a virtual fitting room available in 43 markets.
The common thread: AI is used where it saves time and money, and real people stay in the loop, credited and paid.
3. What went wrong
Not every launch went smoothly. In 2023 Levi's announced it would test AI-generated models to show more body types and skin tones, and was criticised for choosing technology over hiring diverse models. The lesson was clear: AI should not replace representation, it should support a brand's real values.
Shoppers are paying attention too. In a Stylitics consumer survey, 71% of people said they could not tell an AI image from a real photo, and 59% said they want brands to say when an image is generated. Transparency is not a risk, it is part of the brand.
4. What a brand gains
Speed first: a new drop, a new colour or a seasonal update can be ready in days, not weeks. Then consistency: the same face across product pages, campaign, social and newsletter builds recognition, the way a house muse does.
A virtual model also gives range, because one look can be shown in the studio and in the desert, at dawn and at night, without moving a crew. And the budget saved on repeated shoots can go into design, product and the shoots that really need to be real.
5. How to do it well
Start from the product. The garment must stay true in fabric, fit, colour and details: a beautiful image that shows the wrong product creates returns, not sales.
Build the model like a casting, choosing face, age, attitude and styling as you would for a real campaign. The virtual model should look like your brand, not like a generic AI face.
Keep the art direction human. Light, set, references and editing are what turn a generated image into a fashion image, and this is where a creative director makes the difference. Finally, be transparent: label AI images where it matters and respect the rights of any real person involved.
6. How we work at Vertigini Studio
At Vertigini Studio in Milan we create AI virtual models for fashion, beauty and luxury brands: a custom face and body, a full reference set (front, profile, back), and then campaigns, lookbooks and e-commerce images with the same model in every setting. Our own images in this article come from one of our virtual models, shown in the studio and on location.
If you are planning a new collection or a faster e-commerce workflow, take a look at our services or write to us: a short call is the quickest way to see what a virtual model could do for your brand.
Sources: Reuters (May 2025), The Business of Fashion (December 2025), Stylitics consumer survey, announcements from H&M, Mango and Zara.
Comments