When Fashion Nova quietly introduced AI-generated models into its online store earlier this year, the world noticed. At first glance, they appeared like any others: curvy, well-posed and Instagram-polished. But users on TikTok and X pointed out glitchy hands, unrealistic skin textures and a ‘too perfect’ symmetry that gave them away as AI. The virality of these posts sparked a wave of criticism and curiosity over how brands were beginning to replace human talent with AI-generated imagery.
As technology improves, AI models offer brands speed, control and cost savings. But behind the pixel-perfect renderings lie serious legal and ethical questions. Can a virtual model violate a person’s right of publicity? Could it mislead consumers into thinking a real model endorsed a campaign? And what happens to the labour market for human models?
This article examines the growing legal risks fashion brands face as they embrace AI-generated models. As the industry moves into the synthetic era, the law must catch up to ensure that human identity and labour are not overlooked in the name of technological progress.
The Rise of AI-Generated Models in Fashion
The use of AI-generated models is no longer a speculative trend. It is now a shift in how fashion brands produce, promote and present their products. Once reserved for experimental campaigns or digital-only brands, synthetic models are entering the mainstream. High fashion retailers are adopting this technology to optimise visual content at scale.
Platforms like Deep Agency and ZMO.AI offer brands the ability to generate virtual models tailored to specific body types, skin tones and aesthetic preferences. These platforms require no model contracts, no photographers and no production crews. In a matter of minutes, brands can produce dozens of “model shots” for a global customer base.
Retailers such as Zalando have used AI to generate product images for online listings, while Mugler’s Spring 2023 campaign featured a blend of real and digitally enhanced models. Meanwhile, Fashion Nova’s AI rollout in its product photos drew both backlash and fascination from consumers who questioned whether the brand was quietly replacing real women with algorithmic composites.
This movement is not just about aesthetics. AI models are cheaper, available at all times, and immune to labour laws, travel delays or PR scandals. For e-commerce-heavy brands, they allow for instant localisation of campaigns. The same clothing item can be modelled by avatars with different body types or ethnic appearances depending on the target audience. With this convenience comes a redefinition of authenticity.
Right of Publicity: When AI Models Resemble Real People
As AI-generated models grow more sophisticated, they sometimes look a little too real. Whether intentional or incidental, some of these synthetic models bear a striking resemblance to existing models, influencers or celebrities. When that happens, fashion brands could face serious legal exposure under right of publicity laws.
The right of publicity gives individuals legal power to control and profit from the commercial use of their name, image, likeness or other identifiable aspects of their persona. While this right varies by jurisdiction, it is most robust in states like California, New York and Tennessee, and is recognised either statutorily or through common law in many others. Unauthorised commercial use, especially when tied to sales or advertising, can lead to lawsuits, monetary damages and reputational harm.
The risk increases when AI models are trained using publicly available databases that may contain images scraped from the internet. Even if a synthetic model is not a perfect replica of any one individual, the use of facial features, body types or recognisable styles associated with a public figure could be enough to trigger legal action. This was the concern in White v. Samsung, where Vanna White successfully sued Samsung for featuring a robot dressed in a gown posed beside a game board. The image was meant to evoke her persona without using her actual face or name.
More recently, Scarlett Johansson threatened legal action after her voice and image were mimicked in an AI-generated advertisement without permission. This shows that celebrities are increasingly vigilant about the unauthorised use of their likeness in AI contexts. The same concerns apply to the fashion industry, where influencer-driven marketing is the norm and public perception closely ties a model’s identity to a brand’s image.
If an AI-generated model used by a brand bears a close resemblance to a popular Instagram model, especially in the context of a brand that has previously worked with or referenced that model, there may be a case for misappropriation, even if no direct copy was made. The more the AI model suggests a particular person in the eyes of consumers, the more likely a brand could face a claim.
False Endorsement and Lanham Act Claims
Even when an AI-generated model does not explicitly copy a real person’s face, it can still create legal risk if it evokes the appearance or persona of a real-life model, influencer or celebrity. Under the Lanham Act, individuals can bring a false endorsement claim if a brand uses imagery that causes consumers to mistakenly believe the person is associated with or endorses the product.
In fashion, where marketing heavily relies on influencer culture and visual branding, this risk is amplified. If a virtual model closely resembles someone recognisable, consumers might reasonably assume the campaign is tied to that person. Courts have shown that even implied associations can trigger liability when they lead to consumer confusion. This kind of unauthorised connection may be unintentional. If the resemblance is close enough to suggest a real-life figure, the brand could still be held accountable.
Labour Displacement and the Decline of Human Models
As fashion brands adopt AI-generated models, one of the most immediate consequences is the potential displacement of working models. Traditional models, especially emerging and mid-tier talent, rely on steady bookings for income. Replacing them with AI models that do not require pay, travel, breaks or contracts introduces a major cost-cutting incentive for brands and threatens livelihoods in an already precarious industry.
Unlike actors in film or television, fashion models generally lack union protections. This leaves them vulnerable as brands shift to AI models for e-commerce listings, lookbooks and high-end campaigns. If left unchecked, this trend could reduce job opportunities in the modelling industry, particularly for models from underrepresented communities who already face systemic barriers to entry.
While AI tools may help streamline content creation, they raise ethical concerns about creative labour and exploitation. Brands that value authenticity and inclusivity must weigh the benefits of automation against its social and economic costs.
Diversity and Representation Concerns
AI-generated models also pose a risk to the progress fashion has made toward diversity and inclusion. Many AI systems are trained on biased datasets that favour Eurocentric beauty standards, resulting in virtual models that often reinforce narrow ideals of attractiveness.
Even when brands generate models that appear racially or ethnically diverse, doing so without employing real people from those communities raises concerns of performative inclusion. Replacing diverse human models with computer-generated stand-ins undermines representation and risks turning identity into an aesthetic rather than a lived experience.
What Brands Should Be Doing Now
As AI-generated models become more common, fashion brands must tread carefully to avoid legal and reputational fallout. Brands should avoid using training data that includes identifiable individuals without permission, and ensure their models are sufficiently distinct to prevent claims of misappropriation or false endorsement. When there is a chance of consumer confusion, brands should include clear disclosures that the model is AI-generated and not linked to a real person.
To address labour concerns, brands should consider ethical sourcing guidelines for digital talent, as they do for physical supply chains. This could include hybrid campaigns that blend AI with real models or compensating models for the use of their digital likeness when applicable. Additionally, if AI is used to create models meant to represent a specific racial group or body identity, brands should involve real individuals from those communities in the creative process, not just the dataset.
Conclusion
AI-generated models offer fashion brands speed, control and cost efficiency. They also introduce legal and ethical risks that cannot be ignored. From violations of publicity and endorsement rights to the displacement of human workers and the substitution of real diversity with artificial representation, the implications of replacing real faces with synthetic ones are far-reaching.
As the industry embraces innovation, it must also lead with responsibility. Brands that build clear policies, secure appropriate permissions and centre human dignity in their use of AI will not only stay ahead legally, they will also stay relevant in the eyes of increasingly conscious consumers.
Author: Ivie Egharevba | Research: Ivie Egharevba | Editor: Qazi
Sources:
https://law.justia.com/cases/federal/appellate-courts
Second image sourced from The Impression.