top of page
Search

Who Actually Owns AI Art? (Spoiler: It's Complicated)

Writer: Allie McCormack
Allie McCormack
Aug 7
5 min read

So here's a question that's been popping up everywhere lately, whether you're all-in on AI tools or you side-eye them from a safe distance: if an AI generates an image, who owns it?


Is it the person who typed the prompt? The AI company that built the model? Nobody at all? Can a piece of art just... exist, with no legal owner attached to it?


I've been following two real court cases wrestling with exactly this question right now — and they land on some surprisingly different answers depending on how the AI was actually involved in making the thing.


At the heart of it, this really comes down to a question of involvement — how much did a human actually do, versus how much did the AI do on its own? That's the line courts keep circling back to, and it turns out the answer changes a lot depending on where a piece of art falls on that spectrum.


So grab your beverage of choice (iced tea, for me, always), and let's get into it.


Case One: Thaler v. Perlmutter — The "AI Did Everything" Case

Let's start with the extreme end of the spectrum, because it makes the whole "involvement" question really easy to see.


Back in 2018, a computer scientist named Dr. Stephen Thaler tried to copyright a piece of visual art called "A Recent Entrance to Paradise." Here's the twist: he didn't make it. His AI system, called DABUS, did. All of it. He didn't write a prompt, didn't tweak the output, didn't touch it after it was generated. He just listed the AI itself as the sole author and tried to register the copyright that way — basically saying, "the machine made this, and I'd like to own what the machine made."


The U.S. Copyright Office said no. Their reasoning came down to one core idea: copyright law has always required a human author. Not a company, not a machine, not an algorithm — a person. Thaler pushed back and took it to court, and the case worked its way up through the system: the Copyright Office refused it, the district court agreed with the Office, the D.C. Circuit agreed with the district court. Then, this past March (2026), the Supreme Court declined to even hear the case. In legal terms, that means the lower court's ruling stands, permanently, at least for now: a work created entirely by AI, with zero human hand in the process, doesn't get copyright protection in the U.S.


Honestly? This one's not that surprising once you sit with it. Copyright has always protected what courts call "the fruits of intellectual labor" — things that come out of an actual human mind. Ownership, inheritance, the ability to license or sell something, even the intent behind making it — all of that assumes there's a person behind the work. An AI can't want anything. It can't intend anything. It can't own anything either. So legally, it can't be an "author," even if it did all the technical labor of generating the pixels.


Fun (weird) fact: this isn't even the first time a legal fight broke out over a non-human "creator." In Naruto v. Slater, the animal rights group PETA sued on behalf of a monkey named Naruto, who'd snapped a selfie with an unattended camera — arguing the monkey himself should own the copyright to his own photo. Courts ruled the same way there too: no human author, no copyright, no matter how good the shot was.


So Thaler v. Perlmutter answers the easy end of the question: zero human involvement means zero ownership. No person, no copyright, full stop.


But that leaves the much messier, much more common situation completely unanswered: what happens when a person is very much involved — picking every word of every prompt, refining it over and over — but the AI is still the one generating the actual image? That's where things get genuinely uncertain, and that's exactly what Case Two is about.


Case Two: Allen v. Perlmutter — The "But I Worked SO Hard On This" Case


Thaler was the easy case — no human, no copyright, simple as that. Allen v. Perlmutter is the harder one, because this time, a human was very much involved. And that's exactly why it matters: it's the case that forces courts to actually find the line, instead of just confirming that one exists.


Artist Jason Allen created an image called "Théâtre D'opéra Spatial" using Midjourney. And when I say he worked at it, I mean it: he refined the image through more than 600 prompts. Six hundred. This wasn't someone typing one lazy sentence and walking away — this was hours of iteration, adjusting colors, style, composition, tweaking details over and over until it matched the vision in his head. The final piece went on to win first place at the Colorado State Fair's digital art category.


Naturally, Allen tried to register it for copyright. And the Copyright Office said no.


Their reasoning wasn't "you didn't try hard enough" or "you didn't iterate enough." It was something much narrower. They held that no matter how many times Allen refined his prompts, he never actually had control over how the AI interpreted those prompts or generated the resulting image. He could describe what he wanted, over and over, in more and more detail — but the AI was still the one deciding what it actually looked like, pixel by pixel. Six hundred prompts didn't change that underlying fact. Effort and iteration, on their own, just weren't the thing the law was asking about.


Allen pushed back, understandably. His argument is basically: I had a specific artistic vision in my head, and I used Midjourney as a tool to bring that vision into the world, the same way a photographer uses a camera to capture a vision they already had. That comparison isn't random, either — it echoes an old 1884 Supreme Court case (Burrow-Giles Lithographic Co. v. Sarony), which established that a camera operator can be a legitimate author even though a machine is doing the mechanical work of capturing the image.


So where, exactly, is the line between human and AI? That's the million-dollar question, and honestly, nobody has a clean answer yet. This case is still working its way through federal court in Colorado — no final ruling as of this writing — which means the "how much prompting is enough" question is still genuinely open, still being litigated, still very much unresolved. It's one to keep watching.


So... Where Does That Leave Us?


Thaler v. Perlmutter gives us the floor: zero human involvement, zero ownership, no exceptions. Allen v. Perlmutter is still testing the ceiling — how much involvement is actually enough. Until that resolves, the honest answer is: no one knows yet. The courts are still working it out.


Resources:


or... if you REALLY want to get into it:


Thaler v. Perlmutter:


Allen v. Perlmutter:


*A note on the Allen docket link — since the case is still pending, there's no final ruling to link yet; the docket page itself is the best "follow along" resource, since it'll update as new filings happen.


Comments

Rated 0 out of 5 stars.
No ratings yet
Add a rating
Image by Taylor Friehl

Copyright © 2026 Allie McCormack

All Rights Reserved

Website Designed by The Book Khaleesi

bottom of page