Is fashion hurtling towards a big, fat AI problem?

Earlier this month (July 7), Vogue US dropped its August issue. The magazine is fronted by Anne Hathaway, in an editorial photographed by longtime Vogue contributor Annie Leibovitz, with the whole thing a pretty business-as-usual moment from the publication. Lurking inside, however, was a double-page spread that worked social media into a frenzy.

The pages in question were taken over by a Guess advert which, on first glance, didn’t look out of place. The brand’s new-season advert featured two models that epitomised the flawless, all-American beauty it’s been pushing for the best part of 40 years. But the difference between these two women and past muses like Anna Nicole Smith, Gigi Hadid and Sydney Sweeney was that they didn’t actually exist. At the bottom of the page, a tiny credit revealed Guess’s new-season campaign stars were generated by AI.

“When AI is capable of generating ‘perfect’ airbrushed photographs without the need for models, hair stylists, make-up artists, and big production and post-production teams, how long will it be until Vogue and magazines like it are including them in their editorials as a way to cut costs?”

The ad’s inclusion in the magazine marks the first time an AI-generated model has ever appeared in its pages, and it raises many questions about the future of real models working within the fashion industry. When AI is capable of generating ‘perfect’ airbrushed photographs without the need for models, hair stylists, make-up artists, and big pre- and post-production teams, how long will it be until Vogue and magazines like it are including them in their editorials as a way to cut costs?

Beyond this, the ad sets a particularly dangerous precedent when it comes to young, impressionable fashion fans. With AI becoming increasingly hard to spot as the technology advances, rising generations are finding themselves confronted with yet more unattainable beauty standards. Growing up immersed in social media and bombarded by heavily filtered images, a swathe of young adults have been affected by ‘Snapchat dysmorphia’, with surgical cosmetic procedures hitting an all-time high among the demographic. Being confronted by AI models will only exacerbate this.

For a while, fashion was making moves to become more diverse and models of a wider variety of gender, ability, size, and race starring on the covers of glossy mags and bagging big fashion ads. The body positivity movement saw the rise of the curve supermodel, in names like Paloma Elsesser and Precious Lee. But in recent seasons, the industry has taken a massive leap backwards. As per Vogue Business’s seasonal size diversity report, the number of ‘mid’ and plus-size models dipped for the fourth consecutive season for SS25, with larger models making up just 0.3 per cent of those cast in New York, London, Milan and Paris.

From an optimistic point of view, using AI to generate models such as the ones in the Guess ad could open up opportunities to include a wider range of body types in campaigns and editorials – while many companies, casting directors, and stylists would like to diversify their shoots, practical problems like a lack of samples available in bigger sizes thwart their efforts. Though problematic in the way it would cut the amount of jobs for plus-size models working within the industry – many of whom are already struggling to secure bookings due to the about-turn from body positivity back to the super-skinny ideals of the early 00s – some would argue that any representation is still a step forward.

“Even if brands like Guess did want to diversify their models, the menu Seraphinne Vallora serves up – all identikit versions of the same woman with flawless hair and ‘perfect’ bodies – leaves very little option for it to do so”

However, the advert and fallout that followed put a pin in this theory. The AI models were created by Spanish company Seraphinne Vallora, which was founded by former architects turned tech entrepreneurs Valentina Gonzalez and Andrea Petrescu. A quick scroll through the business’s Instagram page reveals image after image of glossy, airbrushed models with flowing hair, button noses, bright eyes, big boobs and tiny waists – and yet Gonzalez and Petrescu insist they are not reinforcing fashion’s narrow, Westernised beauty ideals.

“We don’t create unattainable looks. Actually, the AI model for Guess looks quite realistic,”  Petrescu told the BBC in response to the backlash the campaign has faced. ”Ultimately, all adverts are created to look perfect and usually have supermodels in them, so what we are doing is no different.” However, the pair did admit that their Instagram page was lacking in diversity, but explained that followers “don’t engage much with” posts featuring models with different skin tones. “At the end of the day, we are a business and use images on Instagram that will create a conversation and bring us clients,” they added. Even if Guess did want to diversify its models, the menu Seraphinne Vallora serves up leaves very little option for it to do so.

Even more problematic are their claims to the BBC in regards to creating plus-size models. Petrescu and Gonzalez insist the technology they use is “not advanced enough” to generate models with larger bodies, despite initiatives from the likes of Cat Taylor and Leanne Elliott Young, who have been doing exactly that through their Institute of Digital Fashion platform for the best part of five years proving otherwise. But they also claim that, as the tech improves, it’s down to the brands they work with to change and incorporate different body types into their campaigns – Seraphinne Vallora, like any other company, goes where the money is, providing a product in line with what their client asks for.

For models like Venice Hakeem, the advance in tech is not a welcome development. “As a plus-size model who’s seen opportunities dwindle for us in recent years, this tech is actively erasing us and the decades of progress we fought for,” she says. “It actively replicates and accelerates the erasure we fight against daily, stripping away hard-won visibility and genuine representation. It’s inauthentic and flattens intersectionality, our diverse realities vanish into homogenised beauty, let alone the environmental implications of AI use. It isn’t innovation, it’s simply automating the industry’s worst biases. I believe we deserve better algorithms and better humans.”

“AI actively replicates and accelerates the erasure we fight against daily, stripping away hard-won visibility and genuine representation. It’s inauthentic and flattens intersectionality, our diverse realities vanish into homogenised beauty. It isn’t innovation, it’s simply automating the industry’s worst biases” – Venice Hakeem, plus-size model

The ad isn’t the first time AI models have stirred up controversy. In 2018, the world’s first CGI supermodel Shudu faced similar backlash, with her creator, fashion photographer Cameron-James Wilson, accused of taking jobs away from real-life models of colour. Similarly, back then, social media was awash with comments about what this meant for the future of fashion image-making, with the result… not so much. Shudu and digital influencers like Lil Miquela didn’t go on to make IRL models obsolete (although Lil Miquela did nab a seat at a Prada show), and largely faded into the ether like many other fads. More recently, H&M came under fire for its use of AI models across its e-commerce platform.

Though it’s developing at breakneck speed, AI is still fairly uncharted territory in the fashion sphere. Right now, it’s likely both Vogue and Guess are taking stock of the outcry the AI-generated images have caused and discussing whether taking away the human element from fashion shoots is worth it if it alienates their consumers. The tech might be infiltrating our lives at an unprecented rate, but there are still  many issues that need to be figured out – and the ways in which it is creatively, meaningfully, and ethically utilised in fashion is just one of them.