AI-generated faces are quietly replacing the models Indian fashion promised to hire
Scroll through Instagram tonight, and you will probably stop, mid-swipe, on a face you cannot place. Not because you have forgotten her. Because you never knew her at all.
She has modelled for realme and boAt. She has appeared on Shark Tank India and has over 250,000 followers who comment on her outfits, her skin, her “vibe”. She has never eaten a meal, aged a day, or asked a brand for a fee. She was never in a room. She was rendered.
Her name is Kyra, and she is India’s first virtual influencer. She will not be the last. Somewhere between your last two scrolls, a real 22-year-old in Kolkata or Jaipur may have lost a booking to someone who does not exist.
Models Who Were Never Born
Kyra was built in 2022 by a Bengaluru studio, FUTR Studios. She now fronts campaigns for national brands, was profiled on the digital cover of Travel + Leisure India and later appeared on Shark Tank India when her creators introduced her to investors. She was followed, in the same year, by Naina Avtr, built by Avtr Meta Labs and now branching into her own podcast. Myntra, one of India’s largest fashion e-commerce platforms, went a step further and built its own: Maya, unveiled in 2023 as what the company’s chief marketing officer, Sunder Balasubramanian, called a figure who would “redefine fashion with her trendy and daring approach.” Maya recommends outfits, answers style queries and nudges shoppers towards a purchase twenty-four hours a day, a role once performed by a human stylist.
Among them is the “AI Luxury Saree Queen”, created to sell the idea of Indian tradition without hiring a single Indian to embody it. Each of these personas has, at some point, been marketed as “the first”. Kyra was first. Naina was first. Nearly every launch reaches for the word because, in a market where digital faces can be generated endlessly, being first is the only remaining differentiator. That branding fight is the real story here. It is not about the technology. It is about who gets to claim originality when originality can now be manufactured on a server in Bengaluru or Bengaluru’s Silicon Valley equivalent.
Global Experiment Reaches India
India did not invent this. It imported a script that global fashion had already tested, and the results abroad should worry anyone cheering the trend at home.
In 2023, Levi Strauss & Co. partnered with the Amsterdam studio Lalaland.ai to generate digital models across a wider range of body types, ages and skin tones. The company called it a step towards diversity. The backlash was immediate. New York Magazine branded it “artificial diversity”. Sarah Murray, a working commercial model, told TechCrunch that seeing an AI-generated woman of colour
in a Levi’s denim campaign left her “sad and exhausted”, adding that
modelling was “already challenging enough without having to compete with new digital standards of perfection”. Levi’s walked the initiative back within weeks, insisting it was never meant to replace real diverse casting.
But this was never really about one denim company or one campaign. It was the first public test of a question fashion has now brought to India unasked: does representation require real people, or merely the appearance of them?
The industry did not take the hint. H&M announced in March 2025 it would clone its own models for use across campaigns. Guess ran an AI-generated model in a spread that readers mistook for a real photoshoot. Mango released what it called its first fully AI-generated campaign, placing real garments on synthetic bodies across ninety-five markets. Sara Ziff, founder of the Model Alliance, called
the moment a reckoning: “In an industry that has historically been a
backwater for workers’ rights, this raises critical questions about consent and compensation.” New York’s Fashion Workers Act now requires a model’s consent before her likeness is used to train an AI system, a protection Indian models do not yet have.
Rebecca Valentine of London’s Grey Modelling Agency put it most sharply: “The entire point of diversity representation is to celebrate the authentic, the cultural, the unique.” An algorithm, however well trained, celebrates none of those things. It approximates them.
Notice what none of these companies had to do: hire anyone new. That is the appeal, and the danger, in one sentence.
Even the industry’s own AI evangelists concede the tension. Michael Musandu, founder of Lalaland.ai, built it after growing up never seeing a model who looked like him. He still argues that brands “serious about inclusion will continue to hire” real models of colour alongside digital ones. That is a hope, not a guarantee, and India’s fashion houses have given no indication they intend to honour it.
Globally, the virtual talent pool has professionalised into a marketplace. Lil Miquela has worked with Prada and Calvin Klein. Shudu, billed as the world’s first Black virtual supermodel, has appeared in campaigns for Versace and Balmain. Aitana López, built by a Barcelona agency, earns an estimated €10,000 a month despite
having no pulse. These are not experiments anymore. They are a functioning industry, and India’s Kyra, Naina and Maya are simply the local franchise — proof that whatever happens in this market eventually happens here too.
Sell Diversity, Hire No One
Miuccia Prada once said that what you wear is how you present yourself to the world. Today, the world may be presented by someone who was never born.
Here is the uncomfortable arithmetic. Indian fashion has spent a decade promising better representation: more models from tier two and tier three cities, more body types, more skin tones, more regional faces on national campaigns. That promise required actual hiring. AI has offered brands a shortcut. A platform can now generate the appearance of every size, shade and region it wants, on demand, without booking a flight to Mumbai or a studio in Delhi.
On paper, this looks like the inclusivity fashion always promised. In practice, it lets a brand claim diversity credit while writing zero cheques to the diverse people that credit was supposed to reach. Rajni Daswani of SoCheers says, “These are not just ordinary digital faces. They are controlled, constantly attuned to the brand tone, and can change messaging on the fly.” Unlike a human model, a virtual one cannot go off-script, cannot be “cancelled,” and cannot negotiate a fee.
That control is precisely the problem. A model from Kanpur or Guwahati was never just a face. She was a person whose presence forced an industry to reckon, however reluctantly, with the country it claimed to represent. An AI persona forces nothing. It performs representation without the friction of actually including
anyone.
Imagine opening a bridal lookbook twenty years from now. Every model smiles with a symmetry no face has ever managed to hold for an entire photoshoot. Every drape falls exactly where the software calculated it should fall. Every blush is the shade a thousand data points agreed would sell best. The saree is real. The zari is real. The centuries of craft stitched into it are real. The only thing missing from the frame is a woman who actually wore it.
When Real Models Lose to Fake Ones
Picture a 22-year-old in Kolkata, refreshing her inbox for a reply to a portfolio she sent three weeks ago. Somewhere in the same city, a brand’s marketing team is generating fifty versions of a saree drape on a face that took twelve seconds to render and will never ask about payment terms. She is not competing with another woman. She is competing with a render farm.
This is Myntra’s Maya in miniature: a platform’s own infrastructure, generating the appearance of inclusion at a scale no casting director could match, and at a cost no agency could undercut. It is efficient. It is also a quiet foreclosure. If a brand can manufacture the look of a Kashmiri bride or a Tamil bridal shoot without hiring a single person from either community, the cultures being sold
are extracted, not honoured, and nobody from those communities earns a rupee for supplying the “authenticity” being marketed back to them.
None of this makes the technology itself the villain. Used transparently — disclosed, consented to, paired with real hiring rather than instead of it — AI modelling can cut costs and open doors that a rigid, exclusionary casting system slammed shut for decades. The New York law now forcing explicit consent is proof that regulation, not panic, is the right response. India has no equivalent yet, and
every month without one is a month in which a real model somewhere loses a booking to a face with no name, no rent, and no story of her own.
Fashion has never been only about clothing. It has always been about aspiration, identity and the story of the person wearing the garment. An algorithm can copy the look of that story. It cannot live it. The industry now has to decide, quickly, whether it wants faces that sell clothes or people whose lives give those clothes meaning — because right now, it is choosing the one that never asks to
be paid.
When the model doesn’t exist, the question is no longer whether artificial intelligence can imitate humanity. It is whether humanity is quietly becoming optional.
Top of Form
Bottom of Form
