Trained on You: How Emerging Artists Are Confronting the Machine That Learned From Their Work

Somewhere in the weights and parameters of a large generative model, there is almost certainly something that resembles your work. If you are an artist who has posted images online in the past decade—and nearly every working artist has—the probability is not negligible that your visual language, your compositional instincts, your hard-won aesthetic decisions, contributed in some measure to a system you never agreed to train.

This is the foundational anxiety of the current moment for emerging artists navigating generative AI: not merely that the technology exists, but that their own labor may have helped build it.

A Tool, a Threat, or Both

The artists grappling most honestly with generative AI tend to resist the binary framing that dominates mainstream coverage. It is not simply a tool to be embraced or a threat to be resisted. For many, it is both simultaneously—and the tension between those two realities is where the most interesting creative thinking is currently happening.

Marcella Voss, a 28-year-old illustrator and surface designer based in Portland, Oregon, began experimenting with image-generation platforms in 2022 with what she describes as genuine curiosity. What she found unsettled her in ways she hadn't anticipated.

"I put in prompts that were essentially descriptions of my own work," she says. "Style, subject matter, color palette. The outputs weren't identical to anything I'd made, but they were close enough to be disturbing. It raised this question I couldn't shake: if a machine can approximate my aesthetic based on a text description, what exactly am I selling?"

Voss has since developed a more nuanced relationship with the technology, using it for rapid ideation in early stages while being deliberate about what she considers genuinely her own contribution. But the question she raised—about the market value of a distinctive visual voice in an age of algorithmic replication—remains largely unanswered.

The Dataset Problem

The legal and ethical dimensions of AI training data are currently being contested in courtrooms, legislative chambers, and academic journals simultaneously, with outcomes that remain genuinely uncertain. For emerging artists without the resources to participate in class-action litigation or lobby for regulatory reform, the practical question is more immediate: what can they actually do?

Some are turning to technical countermeasures. Tools like Glaze, developed by researchers at the University of Chicago, allow artists to apply imperceptible perturbations to digital images that disrupt the way AI models process and learn from them. Adoption has been significant among illustrators and digital artists, though the technology's long-term efficacy against increasingly sophisticated models remains an open question.

Others are making deliberate choices about where and how they share their work online—watermarking images more aggressively, limiting high-resolution uploads to platforms with clearer terms of service, or restricting certain bodies of work to physical or password-protected distribution entirely.

"I've become much more strategic about my digital footprint," says Deon Whitaker, a textile artist and printmaker based in Atlanta whose work draws on West African visual traditions. "Not because I'm afraid of exposure, but because I've started thinking about my images as something closer to intellectual property rather than content. That shift in framing changes how I make decisions."

When Artists Choose to Engage

Not every artist's response to generative AI is defensive. A meaningful cohort of emerging practitioners is approaching the technology as a legitimate creative partner—not to replace their own vision, but to extend it in directions that would be impractical or impossible otherwise.

Rafael Montoya, a multimedia artist and recent MFA graduate from the California Institute of the Arts, has built a practice explicitly around the friction between human intentionality and machine output. His recent series, exhibited at a Los Angeles nonprofit space earlier this year, used AI-generated imagery as raw material—deliberately imperfect, uncanny, and productively strange—that he then physically manipulated through painting, collage, and printmaking.

"I'm not interested in AI as a replacement for craft," Montoya explains. "I'm interested in it as a kind of collaborator that doesn't share my assumptions. The outputs are often wrong in ways that are more interesting than being right. That wrongness is where I find the work."

This approach—treating AI's limitations and biases as creative material rather than obstacles—is gaining traction among artists whose practices are already oriented toward process, conceptual rigor, and critical engagement with technology itself.

Authenticity in an Age of Approximation

Perhaps the deepest challenge generative AI poses to emerging artists is not legal or economic but philosophical: the question of what constitutes a distinctive voice when approximations of that voice can be generated on demand.

The art world has long understood that originality is not synonymous with novelty—that artists build on, respond to, and sometimes directly appropriate from those who came before them. But there is a meaningful difference between the human process of influence and synthesis and the computational process of pattern extraction and recombination at scale.

What seems to be emerging, among artists who are thinking carefully about this distinction, is a renewed emphasis on the elements of creative practice that remain genuinely difficult to replicate algorithmically: embodied knowledge, cultural specificity, sustained conceptual development, and the biographical context that gives a body of work its coherence and meaning over time.

"A model can learn to approximate my marks," says Voss. "It can't learn why I make them. And I think that's where the work actually lives—in the why. The job now is to make that legible."

For the generation of artists building their practices in real time alongside these technologies, that articulation—of intention, context, and irreducible humanity—may be the most urgent creative challenge of the decade.