The Courts Just Ruled on AI Art Copyright: What Every AI Creator Needs to Know in 2026
If you make AI art, you have probably spent at least a few sleepless nights wondering whether your creations actually belong to you. Can you sell them? Can someone steal them? What happens if a big company copies your best piece and slaps it on a T-shirt?
Well, the courts have finally weighed in, and the answer is... complicated. On March 2, 2026, the U.S. Supreme Court declined to hear the appeal in Thaler v. Perlmutter, effectively cementing the rule that purely AI-generated works cannot receive copyright protection in the United States. But before you panic and close Midjourney forever, there is a lot of nuance here that actually works in favor of AI creators who know how to play it smart.
Let me walk you through everything that happened, what it actually means for you, and how to protect your work going forward.
The Case That Started It All: Thaler v. Perlmutter
Dr. Stephen Thaler is a computer scientist who built an AI system called the "Creativity Machine." The system generated an artwork titled "A Recent Entrance to Paradise" entirely on its own, without human creative direction. Thaler then filed a copyright registration listing the Creativity Machine as the sole author and himself as the owner through a work-for-hire theory.
The U.S. Copyright Office rejected the registration. Thaler sued. The case wound its way through the federal courts, and the D.C. Circuit Court of Appeals ruled against him. Circuit Judge Patricia A. Millett wrote that "the Creativity Machine cannot be the recognized author of a copyrighted work because the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being."
Thaler appealed to the Supreme Court. On March 2, 2026, the Supreme Court denied certiorari, meaning it refused to hear the case. The lower court ruling stands. AI, on its own, cannot be an author under U.S. copyright law. Period.
What the Ruling Actually Says (and What It Does Not)
Here is the part most people miss when they read the headlines. The court did not say that all AI-assisted artwork is uncopyrightable. It said that a work created solely by an AI system, with no meaningful human creative involvement, cannot be copyrighted.
Judge Millett's opinion actually left the door wide open for human creators who use AI as a tool. The rule "requires only that the author of that work be a human being, the person who created, operated, or used artificial intelligence, and not the machine itself."
That distinction is everything. If you are typing prompts into Midjourney, selecting outputs, editing them in Photoshop, compositing elements, adjusting colors, and making creative decisions throughout the process, you are very likely the author of the final work. The AI is your tool, like a camera or a paintbrush. The Thaler ruling only kills copyright for works where a human presses a button, walks away, and an AI creates something with zero human creative input.
The "Human Authorship" Requirement: How Much Is Enough?
This is the million-dollar question, and unfortunately there is no bright-line answer yet. The U.S. Copyright Office released a major report in January 2025 concluding that works created using generative AI tools can be protected, but only when a human's contribution is "substantial and independently copyrightable."
The Copyright Office has been clear on one thing: prompts alone are not enough. Writing a text prompt is essentially giving instructions, and instructions convey unprotectable ideas, not copyrightable expression. You cannot copyright a recipe's instructions, and you cannot copyright a prompt.
So where does that leave you? The Office says that "what matters is the extent to which the human had creative control over the work's expression." In practical terms, this means the more you do beyond just typing a prompt, the stronger your copyright claim:
Weak copyright claim: Type a prompt, generate an image, save it. That is probably not enough.
Moderate claim: Generate multiple outputs, carefully select the best one, make editing decisions about which elements to keep. Getting warmer.
Strong claim: Use AI to generate raw elements, then extensively composite, edit, paint over, color grade, and arrange those elements into a final artwork with clear creative vision. This is likely copyrightable.
The Copyright Office has already granted registrations for AI-assisted works where the human creator demonstrated significant creative involvement in selecting, arranging, and modifying AI-generated elements. The key case here is Zarya of the Dawn, a graphic novel by Kris Kashtanova that used Midjourney images. The Office granted copyright for the overall selection and arrangement of images and text, but disclaimed protection for the individual AI-generated images themselves.
The Lawsuits Against AI Companies: Where Do They Stand?
While the copyright-for-AI-output question is getting clearer, there is a completely separate legal battle raging over whether AI companies had the right to train on copyrighted artwork in the first place.
The biggest case is Andersen v. Stability AI, a class-action lawsuit filed by artists Sarah Andersen, Kelly McKernan, and Karla Ortiz against Stability AI, Midjourney, DeviantArt, and Runway AI. In August 2025, U.S. District Judge William Orrick ruled that the artists' copyright infringement claims could proceed, and the trial is currently scheduled to begin on September 8, 2026.
Getty Images also has active lawsuits against Stability AI in both the U.S. and UK courts, alleging that Stability scraped over 12 million photographs from Getty's collection without permission to train Stable Diffusion. The UK case is further along, with a trial date expected in late 2026.
These cases matter for AI creators because depending on the outcomes, the tools we all use could face significant restrictions. If the courts rule that training on copyrighted data without permission is infringement, AI companies may need to license training data, which could mean higher subscription costs, different model outputs, or restrictions on certain art styles.
Selling AI Art Commercially: What You Need to Know
Here is the practical question most AI creators ask: can I sell my work?
Yes, you absolutely can sell AI-generated art. Nothing in any law prevents you from selling it. The copyright question is separate from the commercial question. You can sell things you do not hold copyright over. Think of a photograph of the Eiffel Tower at night (the lighting design is copyrighted by the city of Paris, but people still sell photos of it).
The issue is enforcement. If you do not hold copyright, you cannot stop someone else from copying your work. If you post an AI-generated image on Instagram and a company downloads it and puts it on merchandise, you may have no legal recourse if your work is not copyrightable.
This is why the "how much human involvement" question matters so much commercially. The more you edit, composite, and transform AI outputs, the stronger your legal position if someone tries to steal your work.
How to Protect Your AI-Generated Work Right Now
Even without full copyright protection, there are practical steps you can take to protect your AI creations:
Document your creative process. Keep screenshots of your workflow. Save your prompts, your iteration history, your editing steps. If you ever need to demonstrate human authorship, a paper trail of your creative decisions is invaluable.
Add substantial human editing. Do not just generate and post. Paint over elements. Composite multiple generations. Color grade. Add hand-drawn details. Every creative decision you make strengthens your claim.
Watermark and metadata. Embed your name and creation date in the image metadata. Use visible watermarks on preview versions. This does not create legal protection, but it establishes provenance.
Use licensing agreements. If you sell AI art through your own website or platforms, include terms of service that specify how buyers can and cannot use your work. Contract law can protect you even where copyright law falls short.
Consider registering anyway. If your work includes significant human creative input (editing, compositing, arrangement), file a copyright registration. The Copyright Office is accepting AI-assisted works that demonstrate meaningful human authorship. Be transparent about AI involvement in your application, as the Office now requires disclosure.
Stay informed. The Andersen v. Stability AI trial starting in September 2026 could reshape the entire landscape. The Copyright Office is also continuing to release guidance through its ongoing report series on AI and copyright.
What Comes Next for AI Art and the Law
The legal picture is going to keep evolving. The Supreme Court's refusal to hear Thaler v. Perlmutter settles one narrow question (pure AI output is not copyrightable), but leaves the much more important questions about AI-assisted human creativity unresolved at the highest court level.
Neither the Copyright Office, the Patent Office, nor any federal court has drawn a clear line for exactly how much human contribution is "enough" to make an AI-assisted work copyrightable. That line will likely come from future cases, and the Andersen trial in September could provide some of that clarity.
In the meantime, the practical advice is straightforward: use AI as a tool, not a replacement for your creativity. The more you bring your own vision, editing skill, and creative judgment to the process, the stronger your legal position. The law is catching up to the technology, and creators who can demonstrate genuine human artistry in their AI workflows are going to be in the best position when it does.
The era of "type a prompt and call it art" may be legally tenuous. But the era of human artists using AI to amplify their creative vision? That is just getting started, and the law appears ready to protect it.
← Back to Blog