Is Generative Artificial Intelligence the Biggest Art Heist in History?
Generative artificial intelligence showed up promising to revolutionize human creativity, but what a lot of people didn’t expect was the price this revolution would charge the people who spent their entire lives creating. By 2026, it’s pretty easy to see why generative AI has become a problem. The internet itself coined a nickname for what these tools produce: slop — a not-so-flattering term for disposable, soulless content. AI company CEOs take the stage like supervillains, boasting that their products will wipe out entire job categories. The data centers powering these models consume absurd amounts of water and energy. And meanwhile, chatbots around the world fuel delusions, generate misinformation, and have even been linked to tragic cases involving teenagers.
Who predicted all of this before any alarming headlines? The artists. 🎨
Starting in 2022, creators around the world began to notice something strange: distorted versions — almost like cheap knockoffs of their own work — circulating online as if they were original creations. It was no coincidence. Billions of images were scraped from the internet, without notice, without permission, and without any kind of compensation, to feed AI models built by the biggest tech companies on the planet. Artist and writer Molly Crabapple, in a powerful piece published by The Guardian, described the experience of seeing eerie replicas of her own work popping up online — as if made by someone with no talent and under the influence of tranquilizers — with all her lines and textures reduced to mechanical repetition. She didn’t hesitate to call the phenomenon what she considers it to be: the biggest art heist in history.
And who sounded the alarm first? The artists themselves. While big tech executives publicly argued that enforcing copyright laws would kill the AI industry — as venture capitalist Marc Andreessen declared in 2023 — illustrators, designers, and creators watched their markets crumble in real time. The tech industry playbook was well known: move fast and break things. Except the things being broken were real people’s lives and careers. The question hanging in the air, still without a definitive answer, is this: was this just a natural consequence of technological progress, or was it a deliberate choice to ignore the law in the name of profit? In this article, we’re diving deep into this story — from the back rooms of AI companies to the courtrooms, and through the organized resistance of artists who refused to accept the theft in silence. 🔍
How Generative AI Learned to Create — and at Whose Expense
To understand the scale of the problem, you need to take a step back and understand how generative artificial intelligence models actually work. Tools like Midjourney, Stable Diffusion, and DALL-E didn’t come out of nowhere. They were trained on massive datasets composed of hundreds of millions — in some cases billions — of images pulled from the internet. Those datasets included illustrations, photographs, digital paintings, concept art, and any other type of visual content that could be indexed. The problem is that the vast majority of those images had an owner. A real artist, with a name, a portfolio, and years of practice, who never authorized the use of their work for any of this.
The technical process behind Generative AI involves what we call diffusion learning, where the model learns visual patterns from real examples and then recombines those patterns to generate something new. In theory, it sounds harmless. In practice, the result is a machine that can mimic an artist’s style with terrifying precision, using that artist’s very portfolio as its reference — without paying a single cent for it. Platforms like ArtStation and DeviantArt, which for years served as showcases for digital artists, inadvertently became the biggest data sources for training these models.
What makes all of this even more troubling is the fact that the companies behind these tools knew exactly what they were doing. Leaked internal documents, depositions in lawsuits, and public statements from researchers made it clear that the mass data collection was a strategic decision, made with full awareness that copyright-protected content was being included in the datasets. The bet was simple: move fast, launch the product, capture the market, and deal with the legal fallout later — if it ever came.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The Artists’ Pushback and the Resistance Movement
When artists like Greg Rutkowski, Sarah Andersen, Kelly McKernan, and Karla Ortiz discovered that their names were being used as prompts to generate images in artificial intelligence tools, the reaction was shock mixed with fury. Greg Rutkowski, a Polish illustrator known for his epic and detailed style, became one of the most searched names on Stable Diffusion — even surpassing historical references from classical painting. In interviews, he described the feeling as watching his own work being used to build an artificial version of himself, without control, without credit, and without any financial return whatsoever.
The resistance movement didn’t take long to organize. In January 2023, Sarah Andersen, Kelly McKernan, and Karla Ortiz filed a class action lawsuit against Stability AI, Midjourney, and DeviantArt, alleging that the companies violated the rights of millions of artists. The lawsuit, which is still ongoing and contested by the companies, was accompanied by a social media campaign that quickly gained momentum, with artists sharing their work alongside AI-generated imitations to visually demonstrate just how precisely the models could replicate specific styles.
At the same time, artist Molly Crabapple and journalist Marisa Mazria Katz launched an open letter with a straightforward demand: keep AI-generated images out of newsrooms. The initiative attracted thousands of signatures from around the world, becoming one of the defining moments of organized resistance against the unchecked use of generative AI in the media. Crabapple, who had been invited to the Perugia journalism festival in 2023 to speak about using her art to document conflict zones, dedicated a significant portion of her presentation to the threat AI companies posed to creative professionals. She called out how these companies shame their critics, labeling them as stupid and backward, and how the narrative of inevitability is designed to make people passively accept what’s happening. Nothing humans do is inevitable, she said. Everything is determined by politics, money, and power.
Tools like Glaze and Nightshade also emerged as digital defense fronts, developed by researchers at the University of Chicago with a very clear goal: poison AI training data by altering image pixels in ways imperceptible to the human eye but effective enough to confuse the algorithms.
Beyond the legal battles and protection tools, artists began pressuring platforms to adopt clearer policies about the use of content for AI training. ArtStation implemented an opt-out option, allowing artists to flag their work as unavailable for use in datasets. But many critics pointed out that this solution was insufficient, since the damage had been done long before any policy existed, and the models had already been trained on that content. The opt-out logic, instead of opt-in, was widely criticized as shifting the burden of proof: why should creators have to chase down protection for what was already theirs by law?
Big Tech’s Contempt for Human Creativity
As if the unauthorized appropriation of creative work wasn’t enough, public statements from tech industry leaders made the situation even harder to swallow. In 2024, Mira Murati, then OpenAI’s chief technology officer, said in an interview that the creative jobs destroyed by her company’s product maybe shouldn’t have existed in the first place. The statement landed like a slap in the face for millions of professionals around the world.
For Molly Crabapple, that kind of statement reveals the deep anti-humanism of the tech elite. These are people who avoid human interaction — with all its happy coincidences, irritations, and joys — because it represents friction. Learning to make art is also friction. And that friction, Crabapple argues poetically, is the foundation of all pleasure — whether it’s the friction of a pen against paper or the lips of someone you love against yours.
This institutionalized contempt for human creation isn’t just offensive. It carries practical and profound consequences that reach far beyond the art market.
The Legal Battle Over Copyright and AI
At the heart of this entire discussion is a question that legal systems around the world are still trying to answer: does training an artificial intelligence model on copyright-protected works constitute copyright infringement? In the United States, the primary defense used by AI companies has been the concept of fair use, which allows the use of protected material under certain circumstances, such as for educational purposes, criticism, or creative transformation. The argument is that AI training is transformative enough to qualify for this exception. But that argument has not been definitively tested in any major court, and opinions among intellectual property experts are deeply divided.
In Brazil, the situation isn’t much different. Brazil’s Copyright Law — Law No. 9,610 of 1998 — doesn’t specifically address the use of works for AI training, creating a fairly broad gray area. The U.S. Copyright Office published an initial guide on AI and copyright in 2023, making it clear that works generated exclusively by AI are not eligible for copyright protection, since the law requires human authorship. But the inverse question — whether human works can be freely used to train AIs — remains without a clear answer. Meanwhile, lawsuits are piling up in American, European, and other courts, each one trying to establish a precedent that could shape the future of the industry.
What’s at stake goes far beyond a dispute between artists and tech companies. If the courts decide that training on protected data is legal without the need for licensing or compensation, it will create a model where any type of creative content can be absorbed by AI systems with zero return for the original creators. If they decide otherwise, companies will have to renegotiate how their models are built, which could involve massive licensing agreements, changes to datasets, or even retraining entire models from scratch. Neither outcome is simple, and both carry profound implications for the future of the arts and human creativity in the digital age.
The Real Impact on the Creative Job Market
While the lawsuits play out, the impact on the market is already visible and very real. Illustrators who used to freelance for game publishers, animation studios, and advertising agencies report a significant drop in demand for their services since Generative AI tools became mainstream. In specialized forums and creative professional groups, stories of clients who used to pay for custom art switching to AI to produce the same result at a fraction of the cost have become routine.
Crabapple describes this scenario with painful clarity: three years after launching her open letter, AI has torn apart an illustration industry that was already fragile. Many of her colleagues are out of work. And what’s even worse, the entry-level jobs — those projects where young artists used to learn the craft through hands-on experience — have simply been eliminated. The same process is happening across countless creative industries. Human professionals are being replaced by digital homunculi trained on stolen creations. And no, the output isn’t good — but that barely matters. Generative AI works as a tool to discipline and eventually eliminate the human worker. The public will have to get used to it, they say. This is sold as progress.
The Hollywood writers’ and actors’ strike in 2023 had artificial intelligence as one of its central negotiation points, and the outcome was a set of rules that limit but don’t completely ban the use of AI in content production. In the visual arts market, the situation is even murkier, because there’s no collective bargaining agreement strong enough to negotiate these conditions at scale. Each artist, individually, has to decide how to position their work against a technology that was, in a way, fed by that very same work.
What many experts in the creative economy point out is that this isn’t just a short-term problem. If the next generation of creators realizes that there isn’t enough financial return to justify years of dedication to developing a unique artistic style — because that style can be replicated by an AI in seconds — the long-term effect could be a significant reduction in the diversity and quality of human creative output. The irony is that without new human content being created, generative artificial intelligence models themselves will eventually run out of fresh data to learn from, creating a cycle that hurts everyone. 🎭
The Lesson of the Luddites and the Danger of the Inevitability Narrative
When technology advocates want to demonize resistance, they invoke the Luddites. In the version they typically tell, the Luddites were primitive idiots who smashed machines because they were too dumb to understand them. But the real history is quite different. As the book Blood in the Machine by Brian Merchant recounts, the Luddites were skilled artisans fighting for their way of life against textile factories powered by child semi-slave labor. Banned from forming unions, they destroyed machines as a form of protest. And they didn’t lose to the inevitable march of progress. They lost to brute force: the government sent in troops, and the Luddites were executed or deported to penal colonies in Australia.
Artists are also fighting for a way of life. And Crabapple warns that if they’re too disorganized to win, the loss will belong to all of us. The improper data collection by AI companies may have started with the work of illustrators, but it has expanded to encompass everything else. It extends to the billions of dollars these companies waste every year, the carbon they burn, the rare minerals in their chips, the land on which their data centers are built, our culture, our education, our sanity, and our very imaginations. In exchange for all of that, the artist argues, the tech elite can only offer dystopia: a future without meaningful work or real communities, just robots talking to each other with nothing left for us.
What Lies Ahead
The landscape is still taking shape, and the next few years will define much of what happens in the relationship between Generative AI, the arts, and copyright. Some companies have already started pursuing licensing agreements with image agencies and content platforms, as was the case with the partnership between Getty Images and Nvidia to create models trained on licensed images. This move, while still modest, signals that at least part of the industry recognizes that the path of indiscriminate data collection has its limits.
AI-specific legislation is advancing in various parts of the world. The European Union, with its AI Act, includes provisions requiring transparency about the data used to train models, which could force companies to disclose which works were used and open the door for more substantiated copyright claims. In Brazil, bills related to AI regulation are under discussion in Congress, but there’s still no definitive text that specifically addresses the issue of training data and creators’ rights.
Books like Blood in the Machine by Brian Merchant, Enshittification by Cory Doctorow, and Technofeudalism by Yanis Varoufakis help contextualize the moment we’re living through, offering historical and economic perspectives on how technology, when stripped of regulation and checks and balances, can concentrate power and destroy entire ways of life.
What seems certain is that ignoring human creators is not a sustainable strategy for the artificial intelligence industry. Human creativity is both the raw material and the ultimate destination of everything AI tools produce. Finding a model that compensates and respects artists while still allowing technological advancement is the real challenge on the table — and solving it will require far more than better algorithms. It will require political will, corporate transparency, and above all, the recognition that behind every image that fed these models, there was a real person who chose to create. ✍️
