AI Made the Art. But Who Gets the Credit?

When the tool becomes the artist, what happens to the human behind the prompt? The most uncomfortable question in contemporary creativity.

That’s the heart of the current AI‑art debate: if you type a prompt into Midjourney or DALL·E and get a stunning image, who is the artist—the person who wrote the prompt, the team that built and trained the model, or the countless human artists whose work the AI learned from?

What the law says right now

Legally, most major copyright authorities still say: no human, no copyright. In the U.S. “human authorship” is a bedrock requirement, and courts have repeatedly upheld the Copyright Office’s position that a work created entirely by an autonomous AI system cannot be copyrighted.In the well‑known Thaler v. Perlmutter case, a computer scientist tried to register an artwork made solely by his AI system; the courts and the Copyright Office refused, explicitly saying an AI model itself cannot be the author.

At the same time, the Copyright Office has started granting limited protection when humans use AI as one part of a creative process—protecting things like the human‑written text in a comic and the human “selection, coordination, and arrangement” of AI images, but not the raw AI‑generated images themselves.journals.law.Legally, that means the “credit” (in the copyright sense) goes to the human who meaningfully shapes and assembles the work, not to the model and not to the AI output on its own.

What people feel about credit

Psychologically and culturally, it’s messier.An experimental paper literally titled “Who Gets Credit for AI‑Generated Art?” found that people’s sense of authorship shifts depending on how the AI is described: when the system is anthropomorphized—talked about as if it’s an autonomous creator—people are more likely to see the AI, not the human, as the artist.sciencedirect+1Public hype around works like the AI‑generated portrait sold at Christie’s framed the AI as if it alone “created” the artwork, even though multiple humans coded, trained, curated, and marketed the piece, and none of those upstream contributors shared in the fame or the money.

There’s also the invisible crowd: human artists whose images were scraped, often without permission, to train the models that now “paint” in their style.uta+1Experts have warned that museums and platforms now display AI‑made work that clearly riffs on famous artists, without any mention of the originals, leaving viewers with the impression that the AI is the true originator.

The prompter, the model, and the dataset

In everyday creative workflows, at least three human groups lurk behind “AI made the art”:

• Prompt writers / prompters, who iterate prompts, choose seeds and settings, pick from dozens of outputs, and sometimes heavily post‑edit results.

• Model builders, who design architectures, select training data, and define the system’s capabilities and biases.

• Source artists and photographers, whose works are bundled into massive datasets and whose influence lives on in the AI’s “style,” often without consent or attribution.

• 
  

Right now, only the first group—the visible human user—has a straightforward path to legal credit, and even that is limited to cases where they add enough original, human creativity. The model creators are usually recognized in a technical sense (papers, product branding) but rarely credited on each artwork.The dataset artists are typically not credited at all, despite their styles being mimicked, which is why so many working artists see AI art not as neutral “tool use” but as uncredited appropriation at scale.

When the tool starts looking like an artist

Part of what makes this so uncomfortable is that generative models don’t feel like Photoshop brushes—they feel like collaborators that can propose compositions, styles, and ideas the human wouldn’t have invented alone. That tempts platforms and audiences to talk as if the model is the “real” creator, sidelining human labor and erasing the lineage of training data behind each image.

Legally, though, courts and agencies are drawing a hard line: no matter how advanced the model, they treat it as a tool that can’t own rights or claim authorship. The real open question is not whether the AI “deserves” credit, but which humans do—and how much human shaping is enough to count as genuine authorship.

So who should get the credit?

Right now, the emerging compromise looks something like this:

• Credit and rights go to humans who make original creative decisions—prompters, editors, curators, and assemblers—when they contribute more than just hitting “generate.”

• Model creators and dataset artists deserve technical and moral credit, and many experts argue for built‑in provenance tags so AI‑generated work can carry metadata that identifies key source artists and datasets.

• The AI itself remains a powerful, controversial tool—not an author—with no independent claim to legal or moral credit under current frameworks.

  
  

That leaves your core question—“what happens to the human behind the prompt?”—sitting in a live tension. Legally, they’re the only possible “author” in the chain if they add real creativity; ethically, they’re just one of many humans whose labour is woven into each piece. Whether future norms lean more toward crediting the prompter, the upstream artists, or both will decide whether AI becomes a new brush in human hands—or a black box that quietly absorbs and replaces them.