The answer to “can AI-generated content be copyrighted in the United States?” is now as settled as it’s going to get without Congress acting. The Supreme Court denied certiorari in Thaler v. Perlmutter, Case No. 25-449, on March 2, 2026. The case is over.
Dr. Stephen Thaler had argued for years that his DABUS AI system autonomously created a piece of visual art and that the work deserved copyright protection despite having no human author. The Copyright Office denied the registration. The lower courts upheld the denial. The Supreme Court declined to disturb those rulings. Human authorship is a prerequisite for copyright protection. That’s the rule. It was the rule before this case. It remains the rule now, with considerably more legal weight behind it.
The practical implications run in two directions. For developers and companies deploying generative AI tools, fully AI-generated outputs, images, text, audio, don’t carry copyright protection for the deploying organization. That’s not a new exposure. It is, however, a confirmed one. Business models that depend on asserting exclusive copyright over AI-generated content need to be re-examined against this framework. The question is no longer “what might a court decide?” It’s “what does existing, now-reinforced law require?”
The second implication runs toward human-AI collaborative works. Copyright protection remains available where a human author exercises sufficient creative control and selection over the final output. The line isn’t always clean, but it’s the line that matters commercially. Products and workflows designed to keep a human meaningfully in the authorship loop aren’t just good practice – they’re the mechanism for preserving IP value in AI-assisted creative work.
A March 2026 Copyright Office guidance document was reported by multiple outlets but could not be independently verified at time of publication, copyright.gov was inaccessible. The Copyright Office has maintained its longstanding human authorship position consistently through the Thaler litigation. Readers should check copyright.gov directly for the most current official statements. Legal observers note that with legislative action on AI copyright not imminent, courts are likely to resolve the most contested questions around AI and fair use going forward.
What’s next. The AI copyright frontier shifts to fair use. Training data litigation, cases involving whether using copyrighted works to train AI models constitutes infringement, is the next contested terrain. Those cases are working through the courts now. The Thaler ruling doesn’t resolve them, but it does confirm that the judiciary, not Congress, is the venue where AI IP law is being written in real time.
TJS synthesis. If your organization generates content with AI tools, run a quick audit: which outputs does your business model depend on as proprietary? For each one, ask whether a human author exercised sufficient creative control to establish copyright. If the answer is “the AI did most of the work,” that output is in the public domain. Adjust your IP strategy accordingly, and document the human creative contribution wherever it exists, because that documentation will matter if you’re ever challenged.