Last month, I noticed something in a catalogue for a mid-range Australian fashion retailer. The models looked perfect. Not perfect in the way that human models look after retouching — perfect in a way that felt slightly uncanny. The skin was too smooth. The poses were technically correct but lacked the micro-imperfections that make a human body believable in motion.

I checked the credits. No model names listed. No photographer credited for the shoot. Just “images generated using AI technology” in fine print at the bottom of the page.

This isn’t a hypothetical scenario anymore. AI-generated models are being used in Australian advertising right now, and the industry needs to have an honest conversation about what it means.

Where It’s Already Happening

The use of AI-generated imagery in Australian advertising falls into a few categories, and they’re not all equally concerning.

E-commerce product shots. This is the most widespread use. Brands are generating model images for online product listings — showing clothing on a generated body rather than hiring a model for a studio shoot. The cost savings are significant. A traditional e-commerce shoot with a model, photographer, stylist, and studio rental costs $3,000 to $8,000 per day. An AI-generated equivalent costs a fraction of that.

Several Australian e-commerce platforms have quietly begun using these tools. The models aren’t realistic enough for close-up editorial work, but for a thumbnail image on a product page? Most consumers won’t notice the difference.

Social media advertising. Brands running high-volume social media campaigns need constant visual content. Some Australian marketing agencies are now using AI-generated imagery for lower-tier social posts, saving the real model bookings for hero campaigns. One agency I spoke with estimated they’d reduced their model booking costs by 40% in the last twelve months through selective use of generated images.

Virtual influencers. This is the one that gets the most media attention, but it’s actually the least impactful commercially in Australia so far. Virtual influencers like Lil Miquela made headlines globally, but Australian brands have been slow to adopt the concept. A handful of experiments exist, but none have achieved meaningful commercial traction here.

The Economics Are Brutal

Let’s be honest about why brands are turning to AI imagery: it’s cheaper. Dramatically cheaper.

A working model in Australia earns between $250 and $800 for a half-day commercial booking, depending on experience and the agency. Add photographer fees, studio hire, hair and makeup, styling, and post-production, and a single day of content creation costs $5,000 to $15,000 minimum.

AI-generated imagery eliminates almost all of those costs. The tools are improving rapidly — companies like Bria AI and others are specifically marketing to fashion and retail brands, offering generated model imagery that can be customised for skin tone, body type, age, and styling.

For a brand that needs 200 product images a month, the maths is irresistible. That’s the uncomfortable truth.

What This Means for Working Models

The impact isn’t uniform. Different segments of the modelling industry are affected differently.

High fashion and editorial: Largely unaffected for now. The fashion industry values authenticity, personality, and the specific energy a human model brings to a shoot. AI can’t replicate the way a model interprets a designer’s vision in real time, responds to a photographer’s direction, or brings their own emotional presence to an editorial story. Major fashion magazines and luxury brands aren’t going anywhere near AI models — the brand risk is too high.

Runway: Safe. You can’t digitise a catwalk. Australian Fashion Week still requires real humans who can walk, turn, and handle a 30-degree backstage while changing outfits in 45 seconds. No AI can do that.

Commercial print and catalogue: This is where the pressure is real. Catalogue and e-commerce modelling has always been the bread-and-butter income for many working models. It’s the steady work that pays rent between editorial bookings. If brands shift even 30-40% of this work to generated imagery, it represents a serious loss of income for a large number of models.

Content creation and social media: Mixed. Brands still want authentic human content for their primary social channels, but the secondary content — the supplementary images, the variant posts, the volume work — is increasingly being generated. This reduces the total volume of bookings available.

The Diversity Question

Here’s something that bothers me about the AI model conversation, and I don’t hear enough people saying it.

The modelling industry spent the last decade making real progress on diversity. More body types, more skin tones, more ages, more disability representation in casting calls and campaigns. It was slow, it was imperfect, but it was happening because real people demanded it and brands responded.

AI-generated models threaten to undo some of that progress. Not because the technology can’t produce diverse imagery — it can, and that’s part of the marketing pitch. But because the diversity becomes a parameter in a software tool rather than a real person with a real story. There’s a meaningful difference between casting a First Nations model for an Australian campaign and generating an image that looks like one.

The representation matters because real people see themselves reflected in media. When that reflection is a synthetic image rather than a real person who looks like them, the cultural value is different. It’s cosmetic diversity without actual inclusion.

What the Industry Should Do

Australian agencies and industry bodies need to get ahead of this rather than reacting after the damage is done.

Mandatory disclosure. Consumers should know when they’re looking at a generated image versus a real person. The Australian Association of National Advertisers could update their code to require clear labelling of AI-generated imagery in advertising. Some brands are already doing this voluntarily, but it should be standard practice.

Protect the commercial segment. Agencies should negotiate with brands to maintain minimum real-model requirements for campaigns. If a brand wants to use AI for supplementary content, fine — but the hero images, the face of the campaign, the content that defines the brand’s visual identity should feature real people.

Upskill models. The models who’ll survive this shift are the ones who bring something AI can’t replicate: personality, social media presence, creative direction ability, and the capacity to be a brand ambassador rather than just a physical mannequin. Agencies should be investing in developing these skills.

Don’t panic. I know this article sounds alarming, but the full replacement of human models isn’t imminent. The technology has limitations — generated imagery still struggles with hands, complex poses, and the kind of emotional nuance that makes a great fashion photograph. But the trajectory is clear, and pretending it isn’t happening helps nobody.

The Bottom Line

AI-generated models won’t kill the modelling industry. But they will reshape it, and some of that reshaping will be painful for working models who depend on the commercial segment that’s most vulnerable to automation.

The conversation needs to happen now, not in three years when the economics have already shifted. Models, agencies, and brands all have a stake in getting this right. The goal should be finding a balance that takes advantage of new technology without destroying the livelihoods of real people or reducing diversity to a dropdown menu in a software interface.

I’ve been in this industry long enough to know it adapts. But adaptation works best when it’s deliberate, not reactive. It’s time to start being deliberate.