# Who Owns Copyright in AI-Generated Works?

Brooklyn Bishop · October 5, 2026

> Understanding AI Copyright Ownership Under U.S. law, copyright protects human authorship, not an AI model or someone who merely submits its output. The...

## Understanding AI Copyright Ownership

Under U.S. law, copyright protects human authorship, not an AI model or someone who merely submits its output. The Copyright Office and courts reject claims for material generated wholly by an LLM. In Thaler v. Perlmutter, the Supreme Court declined to review the ruling that inventor Stephen Thaler could not claim copyright in a work created autonomously by his system. Prompts, instructions, and repeated attempts usually do not establish authorship, while receiving an AI output creates no automatic right to exclude others from using it.

**Also worth reading:** [Can You Copyright AI-Generated Work When Human Authorship Is Still Required?](https://storywriter.pro/knowledge/can_you_copyright_ai-generated_work_when_human_authorship_is_still_required.php) · [How Can Authors Register AI-Generated Content Under Copyright Law in 2026?](https://storywriter.pro/knowledge/how_can_authors_register_ai-generated_content_under_copyright_law_in_2026.php) · [Who Owns AI-Generated Writing, and What Publishing Rights Should Authors Secure in 2026?](https://storywriter.pro/knowledge/who_owns_ai-generated_writing_and_what_publishing_rights_should_authors_secure_in_2026.php)

Human involvement can change the result. A writer may own copyright in original text, images, music, or code they contribute, and in a sufficiently creative selection, arrangement, or modification of generated material. Publishers should document the human process, review and edit outputs, avoid contractual promises the writer cannot fulfill, and use human-created elements when full protection matters. Vietnam’s recent clarification that qualifying AI-assisted works may receive protection where a natural person makes creative contributions reflects a more flexible approach, but ownership varies by country, so publishing plans should be assessed under the relevant jurisdiction.

## Human Authorship Requirements

In the United States, copyright generally belongs to the human author who creates an original work, not to the AI model, its developer, or the person who merely submits AI-generated text. The Copyright Office and recent judicial treatment—including the Supreme Court’s denial of review in Thaler v. Perlmutter—support the requirement of human authorship. Consequently, a work produced entirely by an LLM may receive no copyright protection, regardless of whether a person typed the prompt or pressed “generate.” Prompting alone ordinarily does not establish authorship.

A person may still own copyright in a larger work containing human-created selection, arrangement, editing, revision, or other original expression. The key question is what the human actually contributed, not whether AI participated. Ownership also depends on employment agreements, work-for-hire status, and assignments; a publisher does not automatically own the author’s rights. Because rules differ internationally, including Vietnam’s emerging treatment of AI-assisted and AI-generated works, creators should document their process, preserve drafts and prompts, and obtain jurisdiction-specific advice before publishing or licensing.

## Supreme Court’s AI Inventorship Ruling

In U.S. copyright law, rights generally belong to the human author of an original work, not simply to the person operating an AI system. The Supreme Court’s denial of review in Thaler v. Perlmutter leaves the D.C. Circuit’s holding intact: an “inventor” must be a natural person, so AI cannot qualify as an author. That does not mean every AI-assisted work is unprotected. Copyright Office guidance says prompts and instructions alone ordinarily do not establish authorship, but substantial human selection, revision, arrangement, or other creative expression may.

A publisher can therefore own copyright in the human-authored elements of an AI-assisted publication. The result depends on what the people contributed and on employment or work-made-for-hire rules; a commissioning contract does not automatically turn a fully machine-made output into copyrighted expression. Purely AI-generated material is not copyrightable in the United States, although parties may agree contractually to use, license, or control it. Companies should preserve drafts, identify human edits, and review outputs for infringement. AI can accelerate production, but it does not replace the legal need for a human creator.

## Reviewing Contracts and Platform Policies

In the United States, copyright generally depends on human authorship. Recent Supreme Court litigation did not resolve who owns AI-generated output; instead, the Court refused to review the D.C. Circuit’s ruling in Thaler v. Perlmutter, which held that works made without a human author cannot receive copyright protection. The practical result is that a work generated entirely by an LLM may have no copyright owner. However, a writer may protect original text, edits, selection, coordination, or arrangement that reflects the writer’s creative contribution.

Ownership also raises contract and policy questions. If AI is used as a tool, the human contributor may own the protectable elements, but a bare prompt is unlikely to establish sufficient authorship. Employment agreements, commissioning contracts, and platform terms can allocate rights between the user, publisher, and AI provider, and may prohibit presenting generated material as fully human-made. Other jurisdictions are beginning to address AI authorship expressly, so publishers should specify disclosure and ownership rules rather than assume one global standard applies.

## Protecting Rights in AI-Assisted Publishing

In the United States, copyright generally belongs to the human author of original expression, not the operator of an LLM. Copyright Office guidance and recent litigation reject AI-only inventorship, so a work generated almost entirely by a prompt may receive no protection. The Supreme Court’s refusal to hear Thaler v. Perlmutter did not rule on the merits, but the human-authorship requirement remains. A claimant must identify original human contributions, such as selection, arrangement, revision, or expressive editing; prompts and acceptance of raw output are generally insufficient alone.

Publishing teams should therefore treat AI output as a draft, preserve prompts and version histories, and document substantial human creative control. Writers may own copyright in the text, structure, characters, and revisions they originate, but should not claim unprotectable machine-generated passages. Employment agreements, commissions, and platform terms can also assign or limit rights. Vietnam’s recent clarification similarly emphasizes that protection turns on human intellectual input, not merely the use of AI. As an AI Publishing Consultant, storywriter.pro can help establish practical workflows for separating protected human work from unsupported output.

## AI Authorship and Copyright Compared

| Situation | Likely copyright position | Practical implication |
| --- | --- | --- |
| Human creates a work independently | The human generally owns copyright, subject to applicable law | Ordinary registration and licensing practices apply |
| AI generates content from a simple prompt | AI-only output may lack copyright protection in the United States | Others may be able to copy the output |
| Human substantially selects, arranges, edits, or transforms AI output | Protectable elements may belong to the human creator | Document meaningful creative contributions and retain drafts |
| AI-assisted publishing in other jurisdictions | Rules vary; some countries, including Vietnam, are clarifying treatment | Obtain local advice before commercial release or registration |

In the United States, courts and the Copyright Office generally require human authorship, while the Supreme Court’s refusal to hear recent AI-authorship litigation leaves existing principles intact. Copyright may protect a person’s original arrangement, selection, editing, or transformation of AI material, but not purely machine-generated expression. Because international rules differ, publishers should document prompts, revisions, and creative decisions.

## Quick answers

### Can an AI legally serve as a copyright author?

U.S. copyright law generally requires a human author, so a work generated entirely by an LLM is typically ineligible for copyright protection.

### Who owns copyright in AI-assisted writing?

A human may own copyright in original expressive elements they create or sufficiently control with AI assistance.

### Does prompting an LLM make the user its author?

Providing prompts does not automatically establish the required human authorship or ownership of the generated output.

### How should publishers document AI contributions?

Publishers should preserve prompts, edits, source materials, model disclosures, and contract terms showing the human creative process.

Canonical: https://storywriter.pro/knowledge/who_owns_copyright_in_ai-generated_works.php
Markdown: https://storywriter.pro/knowledge/who_owns_copyright_in_ai-generated_works.php/index.md