Overproduction is one of the main environmental and economic problems of the fashion industry. And here, AI’s help is particularly needed. The algorithms analyse sales data, buyer behaviour, marketing efficiency and the influence of external factors, such as weather, political and economic conditions, which have become less predictable in recent years. The analysis is used to shape the production and distribution of goods between shops. Studies have shown that such models do not completely eliminate overproduction, but that they significantly reduce surplus stocks and the quantity of goods that are sold [1].

A key technological shift in recent years has been the drift models that make it possible to realistically «receive» clothing for a photograph of a person. Unlike the early 3D examples, the new systems are better able to maintain an invoice for fabrics, dents and logos, folding and landing. Modern studies focus on multi-garment- and multi-view scenarios: when one person is wearing more than one piece of clothing and can be viewed from different perspectives [2]. However, even the most advanced systems do not work well with non-standard bodies, complex materials and traffic. So virtual fitting remains a complement, not a substitute for physical experience.

The Russian market has its own digital models: the Divno brand was the first to use a virtual copy of Julia Brusnikina’s real model for a wedding and cruise collection. It’s a «Ethical AI for e-commerce» briefcase by Tokalon, which triggers digital-twin rocks with real models.

The use of AI not only facilitates the organization of surveys and saves budgets, but makes visible the questions that used to be left behind: who owns the digital body, how its repeated «work» should be paid, and where the boundary between human participation and the autonomous use of its visual presence passes. In changing the usual forms of employment, these practices require a review of human rights approaches to their own image and to the creation of new rules for the payment of filming.

The most common use of LLM — neuronets trained in large volumes of text for understanding, translating and generating human speech — has become marketing texts in product cards, e-mails and advertising banners in fashion — such tools are easily introduced into existing processes, have a rapid and visible effect and can measure results through understandable and measurable metrics. By 2025, however, it had become apparent that automatic generation of text without limitation was blurring brand identity. Companies are therefore moving towards systems that not only take into account legal requirements and tone of change brands, but are gradually developing an individual language that is consistent with DNA and relevant business objectives. AI works here as an internal revision rather than as an autonomous author.

Another important area of application of AI in fashion relates to personal recommendations, not only in online shops, but also in physical rhythm. Modern systems analyse large amounts of data: user-laden visual references, purchase history, type of figure, crop preferences, flowers and styles, and the context of a specific request — for example, a reason or a carrying scenario. On this basis, AI can offer more precise and relevant options than traditional recommendatory algorithms, which rely only on the behavior of similar users.

What is important is that such systems become the basis of truly common experience. By combining online and offline data, AI allows a permanent client to be identified at any stage of contact with the brand and offer him a familiar level of personalization, even in a new store or city. As a result, the recommendations become not just a way to increase sales, but part of a more sustainable relationship between brand and buyer, where attention to identity becomes a key value.

The increase in the use of AI has led to increased regulation. In Europe and the United Kingdom, marking requirements for AI content have already been discussed and introduced, especially in advertising. European AI Act [3] and related codes emphasize the need for transparency: the consumer must understand when dealing with a synthetic image or text. For the fashion industry, where sales are built around building trust in brands, this is particularly important.

AI does not «kill» fashion and does not automatically make it more resilient or more creative. It restructures the infrastructure of the industry by accelerating processes, making hidden steps visible, exposing issues of power, labour and authorship.

Sources