On Monday, March 2, 2026, the US Supreme Court declined — without comment — to hear Stephen Thaler’s appeal in Thaler v. Perlmutter. The denial leaves lower court rulings intact: in the United States, works generated entirely by AI, with no human author, cannot hold a copyright.
Anyone building content products should care. This was the first time the question “can an AI be an author?” reached the Supreme Court’s doorstep. The justices didn’t rule on the merits, but by declining the case they let the existing architecture stand: the Copyright Office’s refusal to register the work, and the district and appeals courts’ unanimous reasoning, now operate as settled national rule.
A Seven-Year Fight Over Authorship
Thaler is a computer scientist who argued that his AI system, the “Creativity Machine,” autonomously generated the 2018 image “A Recent Entrance to Paradise.” He asked the US Copyright Office to register the AI as the author, with himself holding the rights as the machine’s owner. It was a deliberately clean test case: no human prompted the system, no human edited the output, and Thaler refused to substitute himself as the author — he wanted the machine on the certificate. The Office rejected the claim in 2019, and a 2022 review affirmed that the image contains no “human authorship.” In August 2023, Judge Beryl A. Howell of the US District Court in Washington, DC upheld the refusal, producing the most-quoted line of the entire saga: human authorship is “a bedrock requirement of copyright.” A federal appeals court affirmed again in 2025. Thaler petitioned the Supreme Court in October 2025, arguing the rulings had “a chilling effect” on anyone considering creative uses of AI. On March 2, 2026, the Court declined to take the case. It’s over.
Human Authorship: The Bedrock of the System
Place Thaler alongside its neighbors and a consistent through-line emerges: American intellectual property keeps the human at the center of ownership.
The Copyright Office issued guidance in 2025 explaining that AI output produced from text prompts alone is not protected — the human contribution of typing a prompt, in the Office’s view, does not make the prompter an author of what the model returns. The patent side runs parallel: the Federal Circuit has held that AI systems cannot be listed as inventors, and the US Patent and Trademark Office reaffirmed in 2024 that humans may use AI tools in invention but AI itself cannot be the inventor. The UK Supreme Court reached the same conclusion in a separate Thaler case. As The Verge notes, this denial lets “AI cannot be an author” stand on firm ground in US copyright law.
After the Denial: What’s Settled, What’s Not
The settled part is blunt: works generated purely by autonomous AI, with no human author, cannot be registered for copyright in the US. Anyone may copy, reproduce, and commercialize them.
But Thaler was the extreme version of the claim — registering the AI itself as the author. Most real-world output lives in the middle ground: a human writes the prompt, curates outputs, edits, composes, arranges. How much human contribution makes such human-AI collaboration protectable remains a case-by-case question at the Copyright Office, and the Court’s denial does nothing to resolve it. The floor is now certain; the gray zone above it is as gray as ever.
What It Means for Content Teams and Developers
Three concrete consequences. First, purely AI-generated assets — images, illustrations, game art, marketing visuals — sit near the public domain in the US: competitors can copy them legally, and you cannot assert copyright over them. If your product’s visual identity or your game’s art pipeline is raw model output, that output carries no exclusivity. Second, to secure protection you need demonstrable human authorship — editing, selection, arrangement, post-processing — and records of that process, which changes how content pipelines should manage assets: version histories, edit logs, and prompt-plus-revision trails become legal evidence, not just hygiene. Third, stock libraries and platforms that claim to “own” and license purely AI-generated material need to re-examine their foundations; buyers are paying for exclusivity that does not legally exist.
For generative AI products, the implication is uncomfortable but clarifying: the model’s raw output is not a moat. The editing workflow, the curation, the tooling around generation — those are the parts where human authorship can still be claimed, and where defensibility actually lives.
Sources
- US Supreme Court declines to hear dispute over copyrights for AI-generated material — Reuters
- AI-generated art can’t be copyrighted after Supreme Court declines to review the rule — The Verge
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.