The Docket · Legal

Can AI-generated work be copyrighted in the United States?

Ruling current. Adjudicated 20 October 2026; next review by 18 January 2027. 4 of 4 panel models answered.

Not legal or professional advice. This page is informational, not legal, medical or financial advice, nor any other professional advice. It is not a substitute for advice from a qualified professional about your own situation.

On the Surface

Purely AI-generated works cannot be copyrighted in the United States; only the human-authored portions of a human-AI collaboration may receive protection, and then only when human creative choices are substantial and perceptible.

Agreed by all four

  • U.S. copyright law requires human authorship; works created solely by AI without meaningful human creative input are not eligible for copyright protection.
  • The U.S. Copyright Office will register only the human-authored portions of a mixed human/AI work; the AI-generated elements themselves are not protected.
  • A prompt alone is generally insufficient to establish human authorship, because prompting does not give the user sufficient control over the specific expressive content of the output.
  • Substantial human editing, selection, arrangement, or modification of AI output can qualify for copyright protection.
  • Applicants must disclose AI-generated content when registering a work and may claim copyright only in the human-authored parts.

By three

  • In Thaler v. Perlmutter (D.D.C. 2023), a federal court ruled that an image generated entirely by Stephen Thaler's "Creativity Machine" AI lacked the required human authorship and was ineligible for copyright; the D.C. Circuit affirmed in 2025.
  • The Copyright Office's March 2023 guidance states that when AI technology receives a prompt from a human, the traditional elements of authorship are determined by the machine, not the human, so the resulting work lacks the human authorship necessary to support copyright.
  • In the Zarya of the Dawn case (2022
  • 2023), the Copyright Office registered the graphic novel's text and the human author's selection and arrangement of images, but refused protection for the Midjourney-generated images themselves.

By two

  • Works generated entirely by AI are generally considered to enter the public domain upon creation, meaning anyone may use, modify, or sell them without permission or licensing.
  • A sufficiently creative selection, coordination, or arrangement of AI-generated elements by a human can support copyright in that human-authored structure.

Raised by one

  • The U.S. Copyright Office released a formal Copyrightability Report in Part 2 of its AI series (April 2025), continuing to develop policy on AI and copyright.
  • The law in this area is evolving rapidly as new lawsuits are litigated and new regulations are proposed; the human-authorship requirement remains firmly in place as of 2025 but could shift with new legislation or higher-court rulings.
Digging Deeper

WHAT WAS ASKED

Whether AI-generated work can be copyrighted in the United States. No reframe was required. The question is clean.

WHERE THE PANEL DISAGREED

There is no substantive disagreement among the four responses. All four reach the same conclusion: purely AI-generated work is not copyrightable in the United States; human-authored elements within a mixed work may be protected; prompts alone do not constitute sufficient creative authorship.

The differences are presentational. One response introduces the 2025 D.C. Circuit affirmation of Thaler v. Perlmutter, which the others do not mention. One notes explicitly that entirely AI-generated work falls into the public domain upon creation. These are additive observations, not challenges to the shared conclusion.

One response hedges that "the assessment is fact-specific and remains unsettled in some circumstances." This is accurate as a description of edge cases, but it does not produce a competing answer. It is a qualifier on the same answer, not a dissenting answer.

HOW STRONG THE AGREEMENT IS

The agreement is strong, but it carries one structural limitation worth naming.

The piece of evidence that would overturn this conclusion is congressional or court action establishing that AI systems can hold authorship rights, or that a prompt constitutes sufficient creative expression. Neither exists in current U.S. law. The overturning condition is real but not imminent.

The agreement does trace partly to a shared source: all four responses rely on the same U.S. Copyright Office guidance and the same body of case law. That is not a blind spot, those are the controlling authorities. But it means the panel is not four independent perspectives confirming each other from different vantage points. They are four readings of the same record. The record is authoritative. The independence is limited.

The strongest counter-argument the panel did not fully develop: the Copyright Office's "human creative control" standard is not precisely defined, and its application is fact-specific. A highly iterative, directive prompting process, where a human makes hundreds of specific expressive choices, sits in ambiguous territory. The Office has not drawn a bright line between "prompt" and "creative control." This is where future litigation will most likely concentrate, and none of the four responses gave it more than a sentence. It does not change the verdict on the core question, but it matters for anyone navigating the edge.

WHAT THIS MEANS

The core question has one clear answer under current U.S. law. The human authorship requirement is not contested by courts, the Copyright Office, or the relevant statutory authority. The panel's agreement reflects the state of the law, not a shared assumption about what the law should be.

The unsettled territory is not "can AI-generated work be copyrighted", that answer is no. The unsettled territory is where the line falls between a prompt and meaningful human creative control. That line is being drawn case by case, and the Office's guidance is more directional than definitive on the edges.

BOTTOM LINE

Purely AI-generated work cannot be copyrighted in the United States. The U.S. Copyright Office and federal courts, including the D.C. Circuit's 2025 affirmation of Thaler v. Perlmutter, have held that human authorship is a non-negotiable threshold requirement. A prompt submitted to an AI system does not meet that threshold.

Where a human makes substantial, perceptible creative choices, editing, arranging, selecting, or modifying AI output in ways that reflect original expressive judgment, those human-authored elements may qualify for protection. The AI-generated portions do not, regardless of how the work is packaged or registered.

The decisive factor: copyright protection in the U.S. attaches to the author, and authorship requires a human being. That requirement predates AI and has survived every legal challenge to date.

What would change this verdict: federal legislation explicitly extending copyright to AI-generated output, or a Supreme Court ruling redefining authorship. Neither is on the immediate horizon.

The actionable directive: if you are producing work that combines human creative choices with AI-generated material, document the human decisions, what you selected, rejected, modified, and arranged, at the time of creation. That record is what a copyright registration will be built on. Register only the human-authored elements and disclose the AI-generated content to the Copyright Office as required.

Both blocks are SimulQ’s result exactly as a member receives it: four frontier models answered independently, and SimulQ’s protocol compared and ruled on their answers. Nothing above has been edited.

Sources the panel cited

  1. Copyright and Artificial Intelligence | U.S. Copyright Office, copyright.gov. Checked 20 October 2026.
  2. Why the Obsession with Human Creativity? A Comparative Analysis on Copyright Registration of AI-Generated Works - Harvard International Law Journal, journals.law.harvard.edu. Checked 20 October 2026.
  3. Clarifying the Copyrightability of AI-Assisted Works | Foley & Lardner, foley.com. Checked 20 October 2026.
  4. Copyrightability of AI-Generated Works - Artificial Intelligence (AI) and Scholarly Communications - Research Guides at University of Maryland Libraries, lib.guides.umd.edu. Checked 20 October 2026.
  5. In the U.S., can AI-generated text or images be copyrighted? - University of Arizona Libraries, ask.library.arizona.edu. Checked 20 October 2026.

Re-adjudication

Every Docket question goes back to the panel at least every 90 days. If the ruling changes, the change is recorded here.

  • 20 October 2026. Re-adjudicated and republished as SimulQ's complete result (On the Surface and Digging Deeper), replacing the earlier summary.
Your question

Ask your own question — 5 free credits.

Start free trial