AI photography is revolutionizing the creative landscape, enabling photographers to explore possibilities beyond what was once imaginable. It frees artists from traditional constraints. It opens up new avenues for storytelling and artistic expression.
One of the most notable advancements is the ability to create highly customizable digital models. These include AI-generated characters tailored to specific poses, adjusting facial expressions, and mimicking cultural aesthetics.
Brands don’t have to hire models or manage complex production schedules. They no longer need to coordinate locations. As it was said before, they generate digital models tailored to specific poses, looks, and settings, adapted to diverse styles and aesthetics. By using those alterations, it makes it easier for brands to experiment with new concepts. They can do this without the limitations of human availability or budget constraints.
Despite that…
Does AI truly cut it?
Creativity, by nature, thrives on exploration and pushing boundaries. As mentioned before, designers search for models with specific looks or poses and AI delivers the service. Next, those digital models can be placed in various environments without the logistical challenges of a traditional photoshoot.
Nevertheless, the growing reliance on AI in fashion photography raises ethical concerns surrounding representation and inclusivity. Artists concur that digital models offer convenience and creative flexibility, even so causing the risk of reinforcing unrealistic beauty standards and marginalizing real models.
Two – Stage Approach
AI models, almost entirely, are trained on datasets that over-represent Eurocentric beauty standards. This leads to issues inaccurately capturing the complexities of different ethnic features. As a result, AI-generated images showcase mainly a homogenized or exaggerated version of racial identity. And, these images reinforce stereotypes rather than offering authentic representation, pointing out a diversity filter designed into AI, rendering misrepresentation.
Authentic representation goes beyond visual appearance, we know that; it involves cultural awareness, respect, and inclusion of real voices. When brands create AI-generated models of different ethnicities, they do not engage with real people from those backgrounds. This approach risks reducing representation to a superficial marketing tool. It fails to be a meaningful commitment to diversity.
Although using AI, the inclusion of a diverse group of … individuals in the detailed prompt led to an output showing some degree of ethnic diversity, AI still struggles with complex searches, including generating models of individuals in a group of rich people. Interestingly, including a group of in the prompt many AI software, seemed to be a key trigger for activating the diversity filter, but not enough to genuinely show any ethnic diversity.
The question is who benefits from these false representations, since in traditional fashion and modeling industries, factual representation creates economic opportunities for real individuals. Models, photographers, stylists, and makeup artists from marginalized communities build careers and gain visibility based on fair, diverse employability.
But, using AI-generated models eliminates these possibilities. Instead of hiring a real Black model, a brand just generates an AI version that aligns with its aesthetic preferences. This illustrates another example of approaching commodifies diversity while sidelining real people who should profit from increased representation.
The lack of accountability in AI-generated diversity prevails.
Real, breathing models, individuals with a voice—they can speak about their experiences, advocate for social issues, and share their perspectives. AI models, on the other hand, do not have agency. They cannot challenge the brands that use them, nor can they push for better representation or fair treatment. This raises questions about whether AI-driven diversity is truly progressive or merely a way for brands to sidestep responsibility.
Moreover, using a prompt specifically designed to elicit an image of only one model of a person illustrates another unaddressed representation problem. Multiply AI software’s results are laden with biases, reflecting stereotypes, like presenting successful people as white, male, and young, which piece by piece reveals yet other issues—beauty and age privilege.
Mirror Mirror
Beyond tokenism, AI-generated media supports deep-seated biases against older individuals. The fashion and beauty industries, long criticized for promoting youth-centric standards, are now replicating these biases in AI-generated content. This portrayal, together with media presentation, reinforces the idea that beauty, relevance, and success are tied uniquely to attractive youth.
AI-driven beauty algorithms assess attractiveness based frequently on mathematical symmetry, even skin texture, and proportionate facial features. While symmetry has traditionally been associated with aesthetic appeal, the privileging of smooth and uniform appearances results in an implicit devaluation of aging features. Simplifying, AI software used in beauty assessments assigns lower attractiveness scores to older individuals due to the presence of asymmetry and skin texture variations. This bias extends to AI-generated advertising campaigns and virtual influencers, where aging models are either absent or digitally altered to look younger.
Ageism in AI is not just about exclusion—it actively shapes societal attitudes toward the process of aging itself. When brands rely solely on AI-driven content, which excludes aging individuals, they send a message that older demographics are not valued as consumers. What’s more, the aging limit of being perceived as a perfect prince/princess ends at 30, which means further eradicating the representation beyond of legal age.
Furthermore, rather than improving their marketing perspective, AI-driven advertising campaigns recurrently use digital alterations to make older individuals appear younger, making beauty with an expiration date because younger is more preferable, profitable, and more sellable.
2Binary
Whittaker, the founder of Google’s Open Research Group and the M-Lab, boldly asserted that anyone with any intention could have fixed AI! Understanding her speech, she was trying to enlighten us with her theory of the origin of AI, rising from a surveillance business model, which has one goal — to make money, more money, and nothing less than money.
Whittaker’s understandable anger towards Google and everything man-related is giving us not that clear a picture of her allegiance to researching the urgent societal impacts of AI, focused on now rather than an amorphous future and, in the meantime, protesting Google’s military contracts. The latter strike she was involved in has been related to the use of AI in supporting Israel, despite Google’s board of members openly promising that AI tech would never be applied in any military conflict.
And, through her life as an activist, Whittaker emphasizes how AI has been affected by historical biases and marginalization or questionable actions; she was supporting already gathered evidence showing that it’s naturalizing racist and misogynist determinations about people’s place in the world, making assumptions behind the veil of computational sophistication is the cause of all faults within AI, within its, named it bravely, male — based model.
Whittaker expressed loudly her disbelief that anyone would say the harms caused somehow directly by AI software happening right now — which are felt most acutely by people who have been historically minoritized including Black people, women, disabled people, precarious workers, and more — that those harms aren’t existential.
With that being said…
Even when AI attempts to represent non-binary and transgender individuals, it often relies on stereotypes rather than authentic portrayals. Digital avatars, AI-generated fashion models, and deep-learning tools used in media production depict gender diversity in ways that are either exaggerated, inauthentic, or tokenistic. For instance, AI-generated representations of transgender individuals lean towards hyper-feminized or hyper-masculinized versions that do not accurately reflect the broad spectrum of gender identities.
It is worth reminding ourselves that most AI systems are built on binary gender frameworks, categorizing individuals strictly as male or female, and this approach excludes non-binary and transgender identities, reinforcing outdated and, some would say, inaccurate notions of gender. The same AI software tends to misgender individuals, relying on datasets that primarily include cisgender individuals, failing to accommodate the diversity of gender expression and identity.
Virtual influencers and models often default to gendered personas based on traditional expectations, such as female-sounding voices, reinforcing gendered assumptions about servitude and compliance. Meanwhile, AI-generated influencers that are male-presenting often embody traditional masculinity, further entrenching gender norms instead of challenging them.
Furthermore, various brands and platforms use AI-generated non-binary and transgender models to signal progressiveness without actually supporting gender-diverse communities. This performative inclusivity, or as many called it, performative activism, allows companies to be seen as liberal while avoiding real engagement.
It leaves us wondering about…
No. 2 copyright infringement
In the digital world, creativity and technology are more connected than ever. Artificial Intelligence (AI) has completely changed the way people create and consume content. From generating digital art and music to writing stories and designing logos, AI-powered tools have made content creation easier and faster than ever before. Even so, this rapid progress has also led to serious concerns about copyright infringement.
AI’s ability to create content quickly and efficiently seems exciting, but it also brings several risks. While AI developers often claim their models are designed to learn rather than copy, many AI-generated works contain elements that closely resemble copyrighted material.
Copyright laws were designed long before AI became a content creator, so courts and lawmakers are now scrambling to keep up. Since AI isn’t a person, it can’t legally hold copyrights. So, do the rights belong to the programmer, the company that owns the AI, or the user who prompts the AI? So far, courts have ruled that AI-generated work isn’t eligible for copyright protection unless a human significantly alters or contributes to it. Copyright laws generally protect original works of authorship.
But if AI-generated content is based on thousands of pre-existing works, is it truly original? Some argue that AI-generated content is more of a remix than a new creation and what with ethic models generated by white creators?
Some AI developers claim their use of copyrighted material falls under fair use, which allows limited use of copyrighted works for purposes like education, commentary, or research. Still, many artists and musicians argue that training AI on their work without permission goes far beyond fair use.
Real-World Copyright Disputes
Many artists have discovered that AI-generated images closely resemble their work — even down to details like brushstrokes and color palettes. In response, many have called for stronger copyright protections that prevent AI models from using their artwork without consent.
The particular scandal involving using the graphic design for the poster of Hawkeye, without crediting David Aja proves the point. The studio was trying to convince everyone by claiming it was the joint work of AI and their in-house graphic designers. They intentionally obscured the real author of the work, suggesting later that Aja’s work wasn’t that good anyway to use in this production.
Despite the warnings, many creative artists still turn to AI software for writing, painting, and design. Its convenience, speed, and accessibility make it an appealing tool, especially in industries that demand constant innovation and quick content production. But as AI becomes more deeply woven into artistic processes, the real question is: what’s the trade-off?
Sure, AI can boost efficiency, but it also poses a risk to the authenticity and originality that define true artistry. If artists lean too heavily on algorithms, they might unintentionally dilute their unique creative voices in favor of machine-generated output. On top of that, there are serious ethical concerns—copyright infringement, data exploitation, and the overall devaluation of human craftsmanship are just a few of the pressing issues.
Looking ahead, artists need to find a balance between embracing technology and preserving their artistic integrity. Instead of letting AI take over the creative process, they should use it as a tool to enhance their work while staying in control of their vision. In the end, technology should support human creativity, not replace it. If artists don’t stay intentional and critical about how they use AI, they risk letting it reshape art in a way that prioritizes automation over authenticity.
Can We Make Everyone Happy Though?
References:
Technopak (2024). Rise of Virtual Models: A New Challenge for Influencers in 2024. [online] Technopak Advisors. Available at: https://www.technopak.com/rise-of-virtual-models-a-new-challenge-for-influencers-in-2024/ [Accessed 18 Feb. 2025].
Shoaib, M. (2023). AI models are here. Can they actually improve fashion representation? [online] Vogue Business. Available at: https://www.voguebusiness.com/technology/ai-models-are-here-can-they-actually-improve-fashion-representation [Accessed 18 Feb. 2025].
Arrigo, L. (2024). The Future of Fashion: AI Generated Models – Ai Model Agency. [online] Ai Model Agency. Available at: https://aimodelagency.com/the-future-of-fashion-ai-generated-models/ [Accessed 18 Feb. 2025].
The New Yorker (2018). Shudu Gram Is a White Man’s Digital Projection of Real-Life Black Womanhood. [online] The New Yorker. Available at: https://www.newyorker.com/culture/culture-desk/shudu-gram-is-a-white-mans-digital-projection-of-real-life-black-womanhood [Accessed 18 Feb. 2025].
Demopoulos, A. (2023). Computer-generated inclusivity: fashion turns to ‘diverse’ AI models. The Guardian. [online] 3 Apr. Available at: https://www.theguardian.com/fashion/2023/apr/03/ai-virtual-models-fashion-brands [Accessed 18 Feb. 2025].
Parham, J. (2020). TikTok and the Evolution of Digital Blackface. [online] Wired. Available at: https://www.wired.com/story/tiktok-evolution-digital-blackface/ [Accessed 19 Feb. 2025].
Sobande, F. (2021). Spectacularized and Branded Digital (Re)presentations of Black People and Blackness. Television & New Media, [online] 22(2), pp.131–146. doi:https://doi.org/10.1177/1527476420983745.
Barton, R. (2021). Marvel Artist Accuses MCU of Stealing Artwork, Fans Launch Boycott. [online] Inside the Magic. Available at: https://insidethemagic.net/2021/11/dave-aja-hawkeye-artwork-fans-boycott-rwb1/ [Accessed 19 Feb. 2025].
Baum, J. and Villasenor, J. (2024). Rendering misrepresentation: Diversity failures in AI image generation. [online] Brookings. Available at: https://www.brookings.edu/articles/rendering-misrepresentation-diversity-failures-in-ai-image-generation/ [Accessed 21 Feb. 2025].
AccessNow (2021). How AI systems undermine LGBTQ identity. [online] Access Now. Available at: https://www.accessnow.org/how-ai-systems-undermine-lgbtq-identity/ [Accessed 23 Feb. 2025].
Metz, R. (2019). AI software defines people as male or female. That’s a problem. [online] CNN. Available at: https://edition.cnn.com/2019/11/21/tech/ai-gender-recognition-problem/index.html [Accessed 23 Feb. 2025].
Porch Group Media (2024). Consumer Shopping Trends by Generation | Porch Group Media. [online] Porch Group Media. Available at: https://porchgroupmedia.com/blog/generational-consumer-shopping-trends/ [Accessed 23 Feb. 2025].
Mead, K. (2024). The Great Wealth Transfer: Baby Boomers To Pass On $53 Trillion To Their Children By 2045. – Finura. [online] Finura. Available at: https://www.finura.co.uk/news/the-great-wealth-transfer-baby-boomers-to-pass-on-53-trillion-to-their-children-by-2045/ [Accessed 23 Feb. 2025].
Newman-Bremang, K. (2024). Is Artificial Intelligence Really Racist? It’s Time To Confront Our Biggest AI Fears. [online] Refinery29.com. Available at: https://www.refinery29.com/en-gb/artificial-intelligence-black-futurism-racism [Accessed 23 Feb. 2025].
Tucker, I. (2023). Signal’s Meredith Whittaker: ‘These are the people who could actually pause AI if they wanted to’. The Observer. [online] 11 Jun. Available at: https://www.theguardian.com/technology/2023/jun/11/signals-meredith-whittaker-these-are-the-people-who-could-actually-pause-ai-if-they-wanted-to [Accessed 23 Feb. 2025].
Financial Times (2018). Financial Times. [online] Financial Times. Available at: https://www.ft.com/content/799b4fcf-2cf7-41d2-81b4-10d9ecdd83f6 [Accessed 23 Feb. 2025].
BURSZTYNSKY, J. ed., (2023). Why Meredith Whittaker believes AI is ‘replicating patterns of inequality’. [online] FastCompany. Available at: https://fastcompanyme.com/technology/why-meredith-whittaker-believes-ai-is-replicating-patterns-of-inequality/ [Accessed 23 Feb. 2025].
CHAN, W. (2023). Researcher Meredith Whittaker says AI’s biggest risk isn’t ‘consciousness’—it’s the corporations that control them. [online] Available at: https://fastcompanyme.com/technology/researcher-meredith-whittaker-says-ais-biggest-risk-isnt-consciousness-its-the-corporations-that-control-them/ [Accessed 23 Feb. 2025].
About The Author
Wonderful ♥️