I want new jeans, AI wants me to lose weight: how virtual try-on tools are fuelling body image issues

The AI that I know and love loathe is fast flooding our timelines with em dashes, Barbie-fied selfies and Gemma Collins’ imagined Met Gala outfits. For some, it’s useful; for others, dependable; and for the creatives, genuinely unsettling. The latest wave of technological innovation (or so it is framed) has now drifted into fashion retail, with AI try-on tools slotting themselves into the shopping experience. “I spent hours using it,” Andrea, 29, from London, tells me, speaking of her experience with ASOS’s virtual try-on tool (launched February 2026). “I thought I looked great,” she says, “it made me want to buy the clothes… until reality kicked in and I realised 'I don’t look like that'.” With the pure intention of finding new season clothing, Andrea hoped that the new assisted shopping tool would… well, assist her shopping experience. But instead, the outcome was far less useful and arguably, detrimental. “My waist had been cinched, my arms and legs smoothed, and my proportions totally edited. It distorted the view I had of myself,” she tells me. Hearing (and seeing) this from a friend made the issue hard to ignore. And it’s this tension between perception and reality that warrants closer scrutiny of AI try-on tools now embedded in fashion retail. So how exactly are these tools programmed to work, and who is it serving to edit bodies in this way? Fitting rooms go digital Major high street retailers, like ASOS and Zara, are rolling out AI try-on tools as part of a broader shift towards digital fitting rooms and personalised shopping. The promise is convenience: There’s no sticky carpet or chewing gummed curtain rail, harsh lighting that washes you out is nowhere to be found, and above all, the faint smell of sizing anxiety has been left on the high street. Instead, you’re in the comfort of your own home, holding your phone at arm’s length, watching a digital version of yourself flicker into existence on a screen. Hearst Owned Andrea’s original, unedited image “We know customers want the confidence of seeing how something will really look, but don't want to be pushed into doing it one way,” Melissa Lim, Head of Digital Product at ASOS, says in an online statement. “Our hybrid approach meets them where they are, giving everyone a try‑on option that feels right for them.” And there is logic here. Industry estimates suggest around 30–40% of online fashion purchases are returned in Europe, making clothing the highest-return category in e-commerce. As for the environmental impact of returns, it’s not exactly light reading. In the UK alone, they are estimated to generate roughly 750,000 tonnes of CO₂ annually, largely through transport and processing. It also lands at a moment when the high street is under severe pressure. Figures from the British Retail Consortium show that online shopping has, quite understandably, become the path of least resistance, helped along by pandemic-era closures and general reprogramming of how (and where) we shop. And I’ll admit it: I’m exactly the kind of shopper these tools are designed for. If I can scroll, click and have a jacket I spotted on Instagram appear at my door a few days later – all without leaving the comfort of my own home – I’m going to do it. So yes – in theory, this is one of the few potentially useful intersections of fashion and AI. If you can see how something looks on your body before buying it, you might shop with more intention, return fewer items and ultimately, waste less. But if what you’re seeing is not your body, but a slightly 'improved' version of it, are you really shopping more efficiently, or just more aspirationally? And if so, surely the buy-try-return process is only compounded when you realise what you’ve ordered would be perfect… on the AI version of you. The try-on illusion AI in shopping isn’t entirely new. Consumers have long used augmented reality tools to try on lipstick shades through Maybelline’s lipstick filters or virtually fit glasses at places like Specsavers. But fashion is an entirely different ball game because fashion AI does not just place a product onto an image – it has to interpret the body itself. Platforms like COORD and Zara generate entirely new avatars based on a user’s photo, while ASOS’s tool digitally clips clothes onto an uploaded image. But when limbs are obscured or clothing overlaps awkwardly, the system fills in the gaps itself – smoothing arms, narrowing waists and subtly reconstructing body parts. And because it begins with a real photograph of you, those edits still feel real, even when they are AI-generated approximations. According to fashion psychologist Jennifer Heinen, the distortion is not a glitch. “These AI tools are trained on decades of fashion imagery where bodies have already been standardised – predominantly thin, white, young bodies posed in ways that minimise certain shapes and forms.” That said, AI does not train itself. Human beings build the systems, select the data and define what “good” looks like. The bodies being prioritised – slimmer waists, smoother limbs, narrower silhouettes – are not random distortions produced by rogue technology. They reflect aesthetic values already deeply embedded within fashion imagery and retail culture. In this sense, “the AI tool isn’t failing. It’s doing exactly what it was trained to do,” says Heinen. Which, arguably, is the most worrying part of it all. And what’s more, these alterations are just subtle enough to feel plausible, says Heinen. “It’s like FaceTune – if you lighten your teeth with two clicks, it still looks natural. If you do it ten clicks, it looks fake. The problem is the subtle nuance.” A 2022 review published in Adolescent Research Review, examining dozens of studies on social media exposure and body image, found repeated exposure to idealised manipulated imagery consistently increased body dissatisfaction among young women. One study found that teenage girls exposed to retouched Instagram selfies reported lower body image than girls shown the original, unedited versions. Another found that exposure to thin-ideal imagery increased body dissatisfaction and lowered appearance satisfaction among young female participants. This all matters because ASOS’s customer base skews young. And I can’t really talk about Gen Z or younger millennial women as some abstract “group” here because this is the audience I’m scrolling alongside. It’s young women like myself, like my friends, like Andrea, who are already the ones most exposed to, and affected by, the constant low-level churn of appearance online. An old trick, newly coded Fashion has always had a chequered relationship with body image, with brands enlarging garment measurements while shrinking labels to flatter shoppers into feeling small (read: the plague of vanity sizing). Feel good, buy more, is the motto. So are AI tools really any different or simply an extension of that same logic for the digital age? Fashion historian Emma McClendon argues that what’s changed is proximity. “The manipulation is no longer just in the images you consume,” she says. “It’s applied directly to the consumer’s own body in real time.” And when that altered image is your own body – not a model, not an influencer, but you – the effect becomes harder to dismiss as “just advertising”. Heinen agrees, describing this as a form of “identity annihilation”. “When you see yourself, and you are not represented the way you actually look, it literally means you are being told that you are not allowed to exist,” she says. For plus-size consumers in particular, she argues these systems can feel less like personalisation and more like “automated erasure.” “The model is trained predominantly on skinnier women,” she says. And in an era of skinny jabs, this only exacerbates a diet culture where we see a potential future for ourselves, a body which could be us if we were only more disciplined. “You see how your body could be,” Heinen says. “And for people already struggling with body image, that can be brutal.” Once you see it like that, AI try-ons don’t really land as neutral tools. They arrive in a culture that’s already found new ways to recycle the old obsession of making women’s bodies smaller to align with conventional beauty ideals and expectations. Rethinking the mirror The changing room mirror was once accused of making women feel bad about themselves. Now the mirror looks back, studies the data and quietly decides which version of you deserves the outfit more. Save, click, kerching. The real winner, ultimately, is the commercial machine AI is feeding – often at the expense of young women’s confidence. So, what does protection actually look like here when young women are shopping? Do we start with safeguards – like how TikTok banned beauty filters for under-18s? Or is the real issue less about age gates and more about accountability from the brands building these tools in the first place? Or, the bigger question looming: do we just bin the AI try-on altogether and go back to fluorescent lighting and the blunt honesty of the changing room mirror? At this point, it feels like the question matters more than the “reassurance” we’re being sold. Cosmopolitan has contacted ASOS to ask how its virtual try-on tool was trained, whether safeguards exist around body distortion and what measures are in place to prevent idealised rendering. At the time of publication, there has been no response.

I want new jeans, AI wants me to lose weight: how virtual try-on tools are fuelling body image issues

Fashion retail’s new AI try-on tools are the latest way to make us feel bad about our bodies, discovers Lia Mappoura.

Related Stories