US judge rules AI-generated art cannot be copyrighted

If you prompt an AI to generate a masterpiece, do you own it? According to a US federal judge, the answer is still a hard no.

On Friday, US District Court Judge Beryl A. Howell ruled that AI-generated artwork cannot receive copyright protection. The ruling firmly backs the US Copyright Office, establishing that human beings remain an essential part of any valid copyright claim.

Machines cannot be hired hands

The legal battle was brought by Stephen Thaler against the US Copyright Office. Thaler had tried repeatedly to copyright an image generated by his algorithm, dubbed Creativity Machine.

His legal tactic was novel: he attempted to register the image as a work-for-hire to the owner of the machine. Under this theory, the AI algorithm would be listed as the author, while Thaler himself would own the resulting artwork.

The US Copyright Office rejected the application multiple times. After a final rejection last year, Thaler sued, arguing that the refusal was arbitrary and contrary to law. Judge Howell disagreed, stating that copyright has never stretched to protect works absent any guiding human hand.

From monkey selfies to supernatural voices

To anchor her reasoning, Judge Howell cited past copyright precedents, including the infamous monkey selfie case where a non-human animal could not hold copyright.

Interestingly, she contrasted that with an older case where a woman successfully copyrighted a book compiled from notebooks filled with words she believed were dictated by a supernatural voice. The difference came down to human effort in compiling and shaping the output.

Still, Judge Howell conceded that copyright is entering uncharted waters. As artists increasingly rely on generative software, courts will face challenging questions about just how much human input is required to claim protection.

Thaler plans to appeal. His attorney, Ryan Abbot, stated that they respectfully disagree with the court's interpretation, while the Copyright Office praised the decision as correct.

Why it matters

This ruling draws a definitive line in the sand: raw outputs from text-to-image models belong to the public domain, not the prompter. For businesses and creators leaning heavily on automated pipelines, relying on pure AI generation introduces a massive IP vulnerability.

The court took a sensible, conservative stance by affirming that human authorship is a bedrock requirement. It prevents companies from claiming millions of synthetic assets as exclusive corporate property overnight.

However, it leaves a glaring gray area untouched. Because the court only handled an image generated entirely by a machine, it did not define the threshold for mixed human-AI collaboration. Does tweaking a prompt twenty times count as a guiding human hand? What about inpainting or color grading?

With major ongoing battles like the Sarah Silverman lawsuit against OpenAI and Meta, or the GitHub Copilot piracy claims, this decision sets an early precedent. Copyright law will continue to protect human creators first, even as the tools they use become increasingly autonomous.