Kate Heussler is witnessing a concerning shift in the fashion industry.
The professional model has done campaigns for big brands — including David Jones, Rebel Sport and Ben Sherman — and now coaches young people entering the industry.
She fears for those who come in purely because they’re “desperate for validation or fame”, warning younger models are “vulnerable” and “a prime target for being scammed”.
Her worry isn’t just that the young models she coaches may find it harder to get work as global brands including Levi Strauss, Calvin Klein and Tommy Hilfiger use artificial intelligence (AI) generated avatars to showcase their latest fashion.
It’s that without adequate regulation, the rise of “deepfake porn” — where someone can superimpose a person’s face onto sexual images or videos to create realistic content that they have never participated in — could become more entrenched.
“It [falling victim to deepfake porn] can ruin your career, it can ruin your relationship,” she warns.
“When it comes to AI, the type of scam … will be something like an open casting call with a big dangling carrot that could be money, cars, diamonds, whatever it is that brings mostly young girls to these castings.
“And you just have to be mindful of where it is that your name, your likeness, your voice, your images, your video content that either you generate for them or they take of you, is actually being used the way that it’s meant to.
“And that’s where having a working contract is always essential.”
Rapid advances in AI are prompting governments globally to scramble in a race to develop laws governing the use of the technology.
This year, Industry and Science Minister Ed Husic released a consultation paper on measures that could be put in place to ensure AI is used responsibly and safely.
Many have pointed to the EU’s proposed Artificial Intelligence Act as an example on which Australia could model its laws.
The paper doesn’t focus specifically on the fashion industry but examines general fears generative AI could be used to create “deepfakes” – fake images, video or audio that people confuse for real ones – which could influence democratic processes or “cause other deceit”.
The paper suggests many of the risks associated with AI can be covered by existing regulation — including Australian consumer, corporate, criminal, online safety, administrative, copyright, intellectual property and privacy laws — and that any gaps will be replaced by new regulation.
Levi’s faced backlash for using AI models to show ‘diversity’
As policymakers try to navigate new regulations, will AI move too fast for them to keep up?
At the end of 2022, the global apparel market was worth about $US1.53 trillion ($2.42 trillion) and is estimated to reach about $US1.7 trillion by the end of 2023.
Notwithstanding the concerns about AI’s negative consequences, many fashion brands are now embracing it to save costs and drive business growth.
A report by McKinsey suggests that in the next three to five years, generative AI could add $US150 billion, conservatively, and up to $US275 billion to the apparel, fashion, and luxury sectors’ operating profits.
This year Levi’s announced its partnership with an AI company, Lalaland, to introduce AI-generated “diverse” models alongside human models.
The Amsterdam-based AI company, co-founded by Michael Musandu and Ugnius Rimsa, uses artificial intelligence to create “human-like avatars”.
Clothing manufacturers can customise these avatars, altering their physical features, such as skin colour, hairstyle and body shape, and can add their own 3D garments to these digital models to display their latest fashion designs at a far lower cost.
Mr Musandu says it’s not just big global brands that are flocking to AI.
“It can be also brands that are finding it difficult to have a budget just to even shoot — to hire real models, photographers, makeup artists, hair stylists,” he tells ABC News.
“We really help level the playing field by allowing them to compete head-to-head against global brands by using the same type of models.”
He says that unlike traditional photo shoots, their technology can create an AI-based model in about five to seven minutes.
“These are all fully digital processes, so that also helps on that whole sustainable impact by reducing physical samples that are needed in such an early stage,” he says.
“By showing more variations, you actually help the brands at least create more value for more consumers.”
But while there are cost savings to be made, there may be reputational consequences of using fake models.
In March, after Levi’s used a diverse avatar created by Lalaland in its marketing, the company faced social media backlash.
Critics called the decision “lazy” and “racist” and questioned why Levi’s wouldn’t simply hire real models to promote diversity.
In a statement, Levi’s said it wasn’t getting rid of real models altogether.
“We are not scaling back our plans for live photo shoots, the use of live models, or our commitment to working with diverse models,” the company said.
“Authentic storytelling has always been part of how we’ve connected with our fans, and human models and collaborators are core to that experience.”
Startups are saving thousands using artificial intelligence
Julie Stevanja is the cofounder of startup Wrapd, an app that allows consumers to unlock discounts and cashback on more than 700 retail brands, and backs up Mr Musandu’s comments about how AI avatars can help small and start-up businesses.
Ms Stevanja has worked in fashion for a decade and at her previous business did more than 100 traditional photo shoots. But now all the imagery she creates at her new business is through AI, at a fraction of the time and cost.
“When I wanted to do a photo shoot before I would have a whole team involved,” she explains.
“I would have a stylist, I would have somebody casting models, we would have to hire models, we would have to plan all of the clothes and the logistics of you know where we’re going to do that and would have to get permits.
“If we were shooting outdoors, you’d have to get photographer lighting.
“So it was probably a team of eight people. And it would be at least of half-day or full-day shoot itself.
“And then you’ve got post-production and you pay for all of that. And you also have to pay for usage rights. Is it going to go on TV? Is it going to go in print? Is it going to go online?
“Now … I can generate the images I need by myself in an hour or so.”
This has resulted in big cost savings.
“It’s gone from spending $10,000 or more on a photo shoot to $60 [for a monthly subscription],” she says.
Ms Stevanja says at her startup she would not have been able to spend $10,000, so AI didn’t kill jobs, it helped generate them.
‘Less time and wastage’ spent on photo shoots
Some fashion brands haven’t moved completely into AI, but are starting the journey.
JAG, a part of APG & Co (the company behind Sportscraft and Saba), is using digital imagery to replace lengthy photo shoots.
Rather than hiring models per individual shoot, models get paid a “usage fee” for the repeated use of their avatars.
JAG’s head designer Katie von Dadelszen says, “there’s definitely less wastage from overdevelopment and sampling”.
She studied fashion and graduated in 2004, hand-sketching fashion designs at the time.
“Back then they weren’t even teaching Adobe Photoshop or Illustrator, and now look at how the world has transformed,” she says.
The company, she says, is no longer paying money for developing samples in factories that are not getting bought or used.
“In the old-school design world, if you designed a collection then you would get samples made with your factories so that you would have physical samples to show your buying team,” she explains.
“If you create a sample and it doesn’t get bought, you’ve got to pay for the fabric. And it’s not just ordering a couple of meters, it’s ordering 30 to 50 meters. So you’ve got leftover fabrics that aren’t being put against an order. And it’s a lot of cost and a lot of waste.
“We’re using 3D design to build these ranges. And, instead of having a physical garment, I’m actually showing a digital garment that I can spin on a turntable.”
Ms von Dadelszen says it also cuts back on having to do ghosting sessions in a studio.
Ghosting is when clothes advertised in a catalogue appear to be worn by an invisible person. But before the image appears, it requires getting either a person or mannequin to model the clothes.
“Garments would be shot on a mannequin, and then the mannequin would be deleted so that you could see a garment filled out with a figure, but there would be no [human] background,” von Dadelszen explains.
“What we do now, rather than having to ghost in our studio … is we’re actually using the design rendered images that we’ve created and coloured and using them as the ghosting pieces.
“So if you click on our website, the first image you see will be a digital-rendered image instead of an image that’s on a model or ghosted.”
She says, over time, AI may play a bigger part in the fashion industry but she believes models will still need to be paid for the use of their avatars.
Do AI models lack ‘authenticity’ or will they kill jobs?
Others argue avatars cannot match the authenticity and human connection that real models bring.
Monique Jeremiah is a model and the director of Diversity Models agency, which hires models from a variety of ethnic backgrounds to promote diversity.
She says AI models will inevitably take away some jobs, estimating about 20 per cent of the work they do may be lost as well as affecting others involved in lengthy photo shoots, such as photographers, makeup artists and stylists.
But she argues the authenticity that real models bring is what will stand the test of time.
“Levi’s are talking about diversity and sustainability and say that’s what they’re trying to represent, but they’re actually contradicting themselves,” she says.
Her human models “come with personality, they come with emotions, and they also come with a story” and this, she argues, resonates with buyers.
“Diversity is basically appreciating the differences in people and AI is not about people — it’s as simple as that,” Ms Jeremiah says.
Lalaland’s Michael Musandu agrees that real models bring authenticity and that will never be fully replaced.
He says what they are aiming to do is supplement them.
“You need models, you need real people, we definitely need it,” he says.
But Mr Musandu argues the shoot is being “streamlined” with the use of AI because the company can pay a model a fee for use of her image and then reuse that repeatedly rather than having to pay for each shoot.
“We also give you the IP and copyright of the model so that you can use it perpetually,” he says.
Julie Stevanja says that while AI may impact some jobs, her startup still works with influencers to run campaigns on Instagram and TikTok.
“It’s an avenue which will really open up in this new space of AI,” she says.
“AI can create imagery, it can’t create personality, it can’t create a following. There is still space for both of those to coexist.”
She says there’s a lot that needs to be worked out around how AI works for the models and the fashion industry.
“There are some artists who are legitimately unhappy about having their artwork and creatives consumed by the AI models,” Ms Stevanja says.
She says regulation should be brought into place to ensure that artists can opt in rather than having to opt out of having their work and images used by others.
“How we get there is still to be determined, but I think we absolutely will see this tool being used,” she says.
Kate Heussler thinks it’s inevitable that AI will result in job losses.
“I don’t think it will happen within one or two years, but I definitely think that within one or two generations, we’re going to see more industries and jobs disappear completely due to AI,” she says.
For now, she’s helping this generation fight to keep their jobs and their reputation.