Digital doubles of real-life models have come onto the scene—and though they’re a newer example of the myriad ways generative AI systems can be leveraged for e-commerce and retail, they’re already making a splash. H&M conjured mixed feelings from the industry when it announced earlier this year that it had created digital twins of real-life models to be used for product imagery.
The Swedish fast-fashion company isn’t the first to leverage that type of technology; WHP Global’s Anne Klein has done the same for more than a year. Still, now that a mass-market household name has publicly let on that it plans to use tech in this way, it has caused a stir among consumers and industry professionals alike.
But what happens to the creatives who have long been part of capturing brand images when those same companies begin using AI for photoshoots? According to those behind the scenes of today’s brand photography, jobs dwindling remains a major fear.
Sara Ziff, founder and executive director of Model Alliance, an advocacy group focused on protecting models’ rights, said that for those working a variety of jobs related to fashion and apparel campaigns and imagery, the prospect of replacement looms large. “There’s a concern about job replacement, not only for models, but also for photographers, stylists, hair and makeup artists, among others,” she said.
Lindsay Kastuk, a makeup artist who works on brand and editorial-style shoots, said the rise of digitally generated product and campaign photos worries her. “A lot of us have poured so much into this specific industry, and [if we] see it all disappear over saving some money, it would be truly heartbreaking,” Kastuk said.
Kastuk said that, while she has not, in her work, encountered a brand planning to replace creative workers with AI, she feels it could be on the way. To the extent she is financially able, she would avoid working with brands leveraging AI to replace creative teams.
“It’s really difficult for me to stand by and support a company that’s, for a lack of a better word, screwing over colleagues,” Kastuk said. “It’s already a really tough industry post-pandemic, and the economy right now is not great—things have slowed down a bit for a lot of people. [AI] is just another hit—you’re like, ‘I just want to do my job. I just want to enjoy it, and I just want to make sure I can keep a roof over my head.’”
Kastuk has not yet put language around AI into her standing contracts, since so much of what happens on a set is out of her control once the model gets in front of the camera, she said. She doesn’t own the images that come out of the shoot, even though her work is featured within them; if those images are fed to AI models generating digital doubles, those systems could learn from hair and makeup artists’ efforts.
Still, Kastuk hopes that the unique looks that she and her colleagues can provide continue to capture brands’ attention as they consider how to proceed with product imagery. Kastuk said that, if jobs become fewer and further between because of the rise of AI, she would have to pivot into other areas of the makeup industry, like wedding looks, which vary significantly from the editorialized style she’s used to.
Related Stories
The level of concern over job stability in certain sectors of the creative industry seems to differ a bit.
Chelsea Loren, who owns her own freelance photography business focused on brand lifestyle shoots, said that, while she has heard some concern floating about in the industry, she remains confident that her skills will continue to be necessary in the coming months and years, even if some brands begin using AI for product photos.
That, she contends, is because, if a wide swath of brands use AI for their product images, they’ll begin to be indistinguishable from one another. “So many brands are going to veer into using AI, and then everything’s going to start looking the same,” Loren said. “I think people are going to want to have that human touch to it.”
Creating digitally generated product imagery can often save brands time and money, particularly if they would have otherwise shot the photos in a destination location. Loren said that, if brands choose to use AI to create product images or campaigns, she knows it could alter their sense of what budget for human-led shoots look like in the future, which concerns her more than full-on job loss.
To help safeguard her work, Loren said she has updated her standard contract to include specific provisions related to AI. Her newer contracts include language that prevents companies from using her work to train AI models and from using AI to alter or create derivatives of her images without written consent.
Both Loren and Elle Dawson, a professional model with 18 years of experience, said part of the issue with AI-generated images is that the products in the images are also digitally generated, which could mean that the way the item fits or how the material looks could be far different from actuality.
“At the end of the day, especially because I’m working with products and consumers and businesses, you don’t want to give [consumers] a false representation of what the product is,” Loren said.
Dawson said she believes that, if this trend continues, consumers are likely to find that products don’t fit the way they’re advertised. If that’s the case, it could see brands fielding higher return rates, in turn damaging profit margins or dashing progress toward sustainability goals. “I think [brands] are using more AI models because they can make [garments] fit on every single body, versus whenever you put a product on me, you have to fake it in different ways,” she said.
Diversity has also been a core issue surrounding AI-generated product content; in 2023, Levi’s said it would partner with Lalaland to create digitally generated models—albeit ones not directly based on real-life models’ likenesses—to introduce more diversity into campaigns and product photos. The denim brand was quickly met with immense backlash, in part because of that diversity push.
Ziff said she feels it’s a further extension of the fashion and apparel industry’s lack of representation to generate digital imagery of diverse models.
“In an industry that’s historically been discriminatory, creating digital representations of models of different races, ethnicities and so on, rather than hiring and paying a diversity of real models, is concerning,” she told Sourcing Journal.
Dawson agreed with Ziff and said the modeling industry is already exploitative, which concerns her because she believes AI could be yet another way models—particularly women—face manipulation.
“The things that models think are OK to deal with are so extreme. Whenever it comes to such a big problem like [AI]…the only way I can trust that the industry won’t go that direction is by being the person who doesn’t let it,” Dawson said. “There has never been a more important time for platforms to be given to the models who actively use their voice to call things out.”
It remains unclear how much models who have licensed their likeness earn from AI-generated photoshoots, and how that rate compares to what those same models would make if they were on a physical set.
Dawson noted that she would only license her digital likeness if a brand offered her “millions” to do so—alongside the ability to speak out on how she’d use that money for good.
“If you’re going to change the world for the negative, you need to make sure you’re getting the resources and planning to do something that makes up for every job you just got rid of,” she said.
As of yet, though, Dawson has not consented to a digitally generated version of her likeness to be used. She believes her physical presence is far more valuable than a digital double of her body, and she said she doesn’t want to further perpetuate impossible beauty standards by allowing a non-real version of herself to be featured in photos promoting a brand or product.
“The one thing I can guarantee to people is that, when you see an image of me, it is me. I was there. I put in the work. It’s my energy—that can’t be replicated,” she said. Dawson said she isn’t concerned about being replaced—but only because she has so much experience and a reputation that precedes her in the industry. Other models, who are newer, lack that same assurance.
“I’m lucky that I get hired based off of my reputation…but I’ve already seen it affect a lot of models, saying their regular jobs are gone,” Dawson told Sourcing Journal.
Dawson advocates for models on a slew of issues, sometimes in partnership with Ziff and Model Alliance, in an effort to ensure fair working agreements and safe conditions. Model Alliance worked on the New York Fashion Workers Act, which has been passed by the state’s legislature and signed into law by Governor Kathy Hochul. The law, which requires greater transparency around models’ contracts and rights, has provisions in it about the requirements for using models’ digital likenesses.
“Beginning on June 19, the [New York] Fashion Workers Act is going to require that modeling agencies and brands that want to create or use a model’s digital replica obtain the model’s clear, written consent, and [that] modeling agencies will no longer be allowed to hold power of attorney over a model’s digital replica, which gives the model more control over how their image is used in connection with generative AI,” Ziff said.
The Fashion Workers Act is a first-of-its-kind piece of legislation, and while Model Alliance hopes to see other states and countries begin to adopt similar tenets, Ziff said she knows it doesn’t even begin to address another important issue: compensation.
She hopes that, as other states and jurisdictions begin to evaluate workers’ rights as they relate to AI, models and other impacted groups have a seat at the table for the discussions. To her, that perspective supersedes any well-intentioned efforts by legislators and companies to speak for models.
“We have tried to take the approach that we’re not anti tech, we’re anti exploitation. We hope that generative AI could be used in a way that is augmenting rather than replacing jobs, but that’s predicated on workers having leverage to negotiate how it’s used, with meaningful protections in place,” Ziff said. “The workers really need to shape the direction of the industry’s adoption of AI before that path is decided for them.”
This article ran in SJ’s Tech Report. To download the full report, click here.
