Photo provided by DepositPhotos
Purely AI-generated images can't be copyrighted in the U.S. After the Supreme Court declined to take up Thaler v. Perlmutter on March 2, 2026, no one owns them, and that answer isn't going to change anytime soon.
For marketing teams now feeding AI visuals into ads and blog headers, that raises a question most never thought to ask: who actually owns the image in your last campaign?
Supreme Court Lets Stand the Rule That AI Alone Can't Be an Author
On March 2, 2026, the Supreme Court declined to hear Thaler v. Perlmutter, in which an individual sought copyright protection for a piece of art entirely created by AI. The Court turned the appeal away without comment, so it didn't write a sweeping new rule. It let a lower court's decision stand.
People are also reading…
That decision matters. The denial left in place the Copyright Office and D.C. Circuit's refusal to register works created purely by AI. The lower courts had repeatedly leaned on one phrase: that human authorship is a "bedrock requirement of copyright." No human hand, no copyright.
So when you generate an image from a text prompt and nothing else, you can use it. You just can't own it the way you'd own a photo you commissioned.
Copyright Office: Even a Detailed Prompt Doesn't Make You the Author
You might assume a long, carefully tuned prompt earns you authorship. The Copyright Office addressed exactly that.
In its January 29, 2025 report, the Office concluded that "given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output." The detail of the prompt doesn't rescue you.
There's a path to protection, but it takes real human work. A work that incorporates AI-generated content may be eligible for copyright if it involves significant human creative input, like a comic book where a person arranges the images and pairs them with original text. The AI pieces on their own stay unprotected.
For a marketing team running on volume, that's a tall order. Few people are hand-editing every social graphic into a defensible derivative work.
The Training-Data Question Is Still Unsettled
Authorship is only half the worry. The bigger legal question is where these models got their training data.
The scale of that fight came into focus last year. Anthropic agreed to pay $1.5 billion to settle a copyright infringement lawsuit brought by a group of authors, a deal the authors' own settlement motion called the largest publicly reported copyright recovery in history. But for anyone hoping for a clean rulebook, note that it was a settlement, not a verdict. The underlying legal line came from a single district court. On June 23, 2025, Senior U.S. District Judge William Alsup ruled in Bartz v. Anthropic that training on lawfully acquired works can be fair use, while the downloading of pirated copies was not. A useful distinction, but one trial court's view, not a settled appellate standard.
That case was about books. Similar questions hang over the image-generation world, where no case has reached a comparable resolution, so the $1.5 billion figure is a books-domain data point, not a measure of image-generator exposure. The legal framework remains open. A definitive ruling from the Supreme Court or even circuit courts has not yet been issued, leaving the framework uncertain and AI companies exposed to ongoing risk.
So a business using a typical AI image generator faces two open questions at once: “I can't own this, and I'm not fully sure the tool that made it was trained cleanly.”
Public-Domain Libraries Sidestep Both Problems
Put those two problems together and a simpler option starts to look smart. If the worry is about ownership and origin, the cleanest route is an image with no ownership claim attached.
That's the logic behind public-domain image libraries like StockCake, which release their visuals into the public domain. When an image is already free of ownership, the authorship question stops being your problem. You're not defending a copyright or filing a registration, and no creator can surface later claiming you used their work.
Think of it like a house with a clean title versus one with a disputed deed. Only one of those holds up when somebody challenges it.
Certainty Is the New Selling Point for AI Visuals
The instinct after a headline like "Supreme Court rejects AI copyright" is to read it as bad news for AI creativity. It's worth flipping around. The Court didn't ban anything; it clarified what you can and can't own, and in doing so it quietly raised the value of images that sidestep ownership fights altogether.
Marketing has spent two years treating AI visuals as a speed-and-cost play. The legal picture is now adding a third factor: certainty. As trade press and creator newsletters keep circling the AI-copyright story, the brands that come out ahead won't be the ones with the cleverest prompts. They'll be the ones who picked images nobody can take back.
The safest AI image, it turns out, is one nobody can own.
