A perfect smile, a flawless body, the outfit fits like a glove – in the online shop the picture looks normal, after all it is a model. Or maybe not? Such images are no longer created just by real people, but by artificial intelligence (AI). Companies are increasingly relying on AI in online shops – with the promise of speed, efficiency and lower costs. For models worldwide, it is becoming a fundamental question: Will the industry even need real people in front of the camera in the future?
AI in the everyday life of international models
“AI is no longer a topic of the future. It is already part of our everyday work,” Ukrainian model Stasya tells the German Press Agency. She has been working internationally as a model for several years, most recently in China for a long time. There she experienced the AI change particularly clearly: faces were digitally transferred to other bodies after shoots. Working hours have also been shortened – from their point of view a consequence of the use of AI. This raises serious ethical and legal questions.
In addition, clients have now submitted contracts that expressly allow the use of model faces for AI purposes. “You think about job security and whether you can still earn enough in the future,” she reports. At the same time, she sees a possible shift in the market: AI could accelerate standardized mass production, while high-quality campaigns with real teams and real sets could gain exclusivity.
Diversity: Real diversity already exists
Berlin model Annika Schnauffer does not see AI as a dominant threat. In any case, competitive pressure is part of an overcrowded industry. Whether an order is lost due to AI usually remains unclear. She could best imagine digital figures in standardized online retail, where large quantities of images are produced with little creative scope. She is critical of the diversity debate: AI can simulate diversity, but “there are enough people and models that you could actually book.”
AI has already arrived in German online retail
In Germany, AI models are no longer a dream of the future. The Hamburg mail order company Otto says it is already using generative AI to create photorealistic model images of synthetic, fictional people. New collections could be available online within a few hours. According to the company, use is limited to the online shop and only takes place during product presentations in which no real people were previously visible anyway. Above all, AI closes those gaps in which classic photo production reaches its economic or organizational limits.
The online fashion retailer Zalando takes a different approach. The company says it works with digital twins of real models. These high-resolution 3D replicas are based on classic photo shoots. Real people would continue to be in front of the camera for creative campaigns. The company says it only uses purely AI-generated models without a real model in exceptional cases – for example, to avoid shoots with children or to react to trends at short notice.
Model agencies between defense and adaptation
In modeling agencies, the debate runs parallel – less about technology than about rights. “We are currently not feeling a massive loss of bookings,” says Carlos Streil, model scout at the Frankfurt model agency East West Models. However, there are inquiries – including from major customers – with contracts in which models should assign rights to AI use, for example to collect data and later use it as an AI model. Such agreements were rejected. In principle, however, people are not against AI as long as avatars are created based on real models and the models continue to earn money.
Carmen Weigel, head of the Berlin modeling agency BMM, argues similarly pragmatically. She believes Zalando’s approach has a promising future: digital avatars of real models could be licensed without the models having to physically appear on set. This is economically attractive, as license fees are often higher than traditional daily fees. However, they do not expect a widespread displacement of real people. For Weigel, modeling is more than photogenicity: “For me, a model is more of a talent,” she says. It’s not just about being beautiful, but about charisma, attitude and identification potential.
Legal: labeling, consent, guidelines
There is also a lot that is still unclear from a legal perspective – especially the question of how AI-supported content must be labeled in online retail. Weigel calls for clear labeling and warns against new, even more unrealistic body images – with consequences for models and young women who would end up comparing themselves to computers.
Even if it is difficult to regulate legally from their point of view, agencies must be particularly careful when licensing images of their models for AI avatars that can be resold to brands: Models must understand exactly what is happening to their face and be adequately compensated for it. Weigel also points to the long-term dimension: once a face has been digitized, it can continue to be used long after death; Informed consent and clear contractual boundaries are all the more important.
Otto describes the debate in Germany and the EU as “in flux”. Binding requirements and guidelines for online trading have not yet been conclusively clarified. Although Zalando emphasizes its claim to transparency, it still sees gray areas: When it comes to digital twins that are based on real models and real clothing and where AI only supports parts of the production, the label “AI-generated” sometimes falls short.
The customers decide
Ultimately, it’s not just the technology that decides in the industry, but the market, says Streil. Weigel also says: “In the end, customers vote with their wallets.” Brands have already experienced how public pressure can change strategies. Weigel refers to “Victoria’s Secret”: After criticism of a narrow ideal of beauty, customers turned away in the 2010s, sales figures fell, and the image began to falter.
It remains to be seen how the industry will develop further. For Schnauffer, this is also a question of power: “In the end, as a model, you are a bit at the mercy of what happens in the industry.” What will likely be crucial is whether clear rules will be enforced: for consent, use and compensation when faces become data.
