The Supreme Court Just Ruled on AI Art. Here’s What It Means for You.
On March 2, 2026, the U.S. Supreme Court made a quiet decision that will shape the future of creativity for decades. It declined to hear the case of Stephen Thaler, a computer scientist who wanted copyright protection for a piece of visual art created entirely by his AI system called DABUS.
The result: AI-generated art cannot be copyrighted. Only humans can be authors.
This ruling didn’t make front-page headlines the way it should have. Because while it settles one question, it opens a dozen more that affect every person who uses AI creative tools, and every artist whose work might have been used to train them.
What the Ruling Actually Says
The case, Thaler v. Perlmutter, has been winding through courts since 2019. Stephen Thaler built an AI system and asked the U.S. Copyright Office to register a piece of art it produced. He listed the AI as the author.
The Copyright Office said no. A federal court agreed. The D.C. Circuit Court of Appeals agreed. And now the Supreme Court has declined to take it up, which means the lower court ruling stands.
The Copyright Office put it plainly: “No matter how many times a prompt is revised and resubmitted, the final output reflects the user’s acceptance of the AI system’s interpretation, rather than authorship.”
In other words, typing a prompt into Midjourney or DALL-E doesn’t make you the author. The AI isn’t the author either. Nobody is. The image has no copyright owner.
This is a bigger deal than it sounds.
What This Means If You Use AI Art Tools
If you’ve been generating images with AI tools for your business, social media, or creative projects, you need to understand what this ruling implies.
Anything generated purely by AI has no copyright protection. That means:
- Anyone can copy it, sell it, or modify it
- You can’t sue someone who uses your AI-generated images
- Your competitors can take your AI marketing visuals and use them freely
- Stock photo sites can reject AI-generated submissions (many already do)
But there’s a gray area. If you use AI as part of a larger creative process, where you’re making significant creative decisions, selecting elements, arranging compositions, and adding your own modifications, the resulting work may qualify for copyright. The Copyright Office has said it evaluates these on a case-by-case basis.
The key question is: did a human make the creative choices, or did the machine?
The $150,000-Per-Song Lawsuit Against AI Music
While the Supreme Court was settling the ownership question, a much bigger battle is playing out in federal courts over how AI gets trained in the first place.
Major record labels, including Universal Music Group, Sony, and Warner, are suing AI music companies Suno and Udio for copyright infringement. The labels claim these companies trained their music-generating models on copyrighted songs without permission or payment.
The numbers are staggering. The labels are demanding $150,000 for every single copyrighted song used in training. If the courts agree, the damages could reach into the billions.
Suno’s defense? Fair use. The company argues that training AI on music is transformative, similar to how human musicians learn by listening to thousands of songs.
But the labels have a devastating counter-argument. They allege Suno used code to stream-rip music directly from YouTube, bypassing anti-circumvention protections. If true, that’s a separate violation that fair use doesn’t cover.
In November 2025, Warner Music Group settled with Suno separately. The terms? Suno agreed to build an entirely new model trained on licensed music and phase out its current models. That settlement signals where this is heading: AI companies will eventually need to pay for training data.
Artists Are Fighting Back (And Winning Ground)
The highest-profile case may be Andersen v. Stability AI, a class-action lawsuit filed by cartoonist Sarah Andersen and other visual artists against the makers of Stable Diffusion.
Their claim: Stability AI scraped 5 billion images from the internet (via the LAION dataset) to train its image generator, including the artists’ copyrighted work. The artists discovered that AI could recreate their distinctive styles simply by using their names as prompts.
In August 2024, Judge William Orrick rejected Stability AI’s motion to dismiss. He found it plausible that “the AI models themselves embody transformations of the plaintiffs’ works.” That case is heading to trial on September 8, 2026.
If the artists win, every AI image generator trained on scraped internet data could be liable.
The Three Battles Happening at Once
What makes this moment so confusing is that three separate legal questions are colliding:
1. Can AI-generated art be copyrighted? The Supreme Court just said no (at least when AI acts autonomously). Human involvement is required.
2. Is it legal to train AI on copyrighted work? This is still being fought in courts. The fair use argument hasn’t been definitively tested yet. The Andersen v. Stability AI trial in September 2026 could set major precedent.
3. Who’s liable when AI copies someone’s style? Even if training is ruled legal, the outputs might still infringe. If you type “paint me something in the style of Sarah Andersen” and the AI produces something recognizably similar, that might be infringement regardless of how the model was trained.
What You Should Actually Do
If you use AI creative tools, here’s practical advice based on where the law stands right now:
For business owners and marketers: - Don’t rely exclusively on AI-generated images for branding or materials you need to protect - Use AI as a starting point, then add significant human creative input - Keep records of your creative process (edits, selections, modifications) in case you ever need to prove authorship - Avoid using artist names in prompts, especially for commercial work
For artists: - Your existing copyright protections are strong. The Supreme Court ruling actually helps artists by denying copyright to pure AI outputs - If you discover AI models can replicate your style by name, document it - Watch the Andersen v. Stability AI trial closely. A win there could establish your right to compensation when your work is used for training - Consider registering your most important works if you haven’t already
For AI music creators: - The legal landscape for AI-generated music is the most volatile - Avoid naming specific artists when generating music - Review the terms of service for any AI music platform carefully - Don’t assume AI-generated music is safe to distribute commercially
The Bigger Picture
What’s remarkable about March 2026 is that we’re watching copyright law get rewritten in real time. The Supreme Court ruling isn’t an ending. It’s one answer to one question in a much larger conversation.
The September trial in Andersen v. Stability AI will likely be the most important technology case of the year. If the court rules that training AI on copyrighted images is infringement, it could reshape the entire AI industry. Companies like OpenAI, Google, and Adobe would face enormous pressure to license training data or prove they have rights to use it.
If it rules the other way, and says training is fair use, artists will have lost a major battle. But even then, the fight over AI outputs that look too much like specific artists’ work will continue.
Either way, the days of AI companies training on the entire internet without asking permission are numbered. Warner’s settlement with Suno is the first domino. Others will follow.
The question isn’t whether AI companies will start paying for training data. It’s when, and how much.
This article is for informational purposes only and does not constitute legal advice. If you have specific copyright concerns about AI-generated content, consult a qualified attorney.