Yes, you can copyright a novel you wrote with AI help, but only the parts a human wrote. Under US law in 2026, your own sentences, scenes and creative choices are protected even if you used AI to brainstorm, outline or edit. Text an AI generated from your prompts is not protected, and if you register the book, you must disclose and exclude any AI-generated material that is more than trivial.
That answer comes from three places: the US Copyright Office's guidance and reports, its registration decisions, and the federal courts. Here is what each says, quoted where the wording matters, and what it means for a working novelist.
First, a quick refresher on how copyright works
Copyright in the US does not depend on registering anything. The Copyright Office's Circular 1 says protection "exists automatically from the moment the original work of authorship is fixed." The moment your draft exists on a page or a file, the human-written expression in it is protected.
Registration matters for enforcement. The same circular says that "Before an infringement suit may be filed in court, registration (or refusal) is necessary for U.S. works," and registering before infringement (or within three months of publication) makes you eligible for statutory damages and attorneys' fees. So the AI question shows up in two places: what is protected at all, and what you have to say when you register.
What the Copyright Office said in January 2025
The Office's most complete statement is Part 2 of its report on copyright and AI, released on January 29, 2025. Its conclusions are short enough to list. These are the ones that matter most for fiction:
- Assistive use is fine. Using AI tools "to assist rather than stand in for human creativity does not affect the availability of copyright protection for the output."
- Your expression stays yours in a mixed work. "Copyright protects the original expression in a work created by a human author, even if the work also includes AI-generated material."
- Pure AI output is not protected. Copyright "does not extend to purely AI-generated material, or material where there is insufficient human control over the expressive elements."
- Prompts are not enough. "Based on the functioning of current generally available technology, prompts do not alone provide sufficient control."
- Selection, arrangement and modification can count. Authors can claim their expression "perceptible in AI-generated outputs," and "the creative selection, coordination, or arrangement" of material, or "creative modifications of the outputs."
- Case by case. Whether human contributions are enough "must be analyzed on a case-by-case basis."
- No extra protection for AI output. The Office found "the case has not been made" for extra or special protection for AI-generated content.
Why prompts don't make you the author
The report explains that prompts "essentially function as instructions that convey unprotectible ideas." Detailed prompts may describe what you want, but "at present they do not control how the AI system processes them." The Office's 2023 guidance gives a writer's example. Ask a system to "write a poem about copyright law in the style of William Shakespeare" and "the technology will decide the rhyming pattern, the words in each line, and the structure of the text." In fiction terms, the model chose the sentences, so the model's sentences are not yours, however good your prompt was.
Brainstorming and outlining with AI
This is the part most novelists care about, and the report is direct. It describes writers using AI as a brainstorming tool, and notes one stakeholder's point that AI "can be used to structure or create a preliminary outline for literary works." Where a writer is "referencing, but not incorporating, the output," the Office says "Using AI in this way should not affect the copyrightability of the resulting human-authored work."
If you asked a chatbot for twenty possible plot twists, picked one, and wrote the chapter yourself, the chapter is yours.
When AI changes text you wrote
The report also covers "expressive inputs," where your own work goes into the tool and survives in the output. One example in it is close to home for novelists: "uploading a story written in the first person and instructing the system to convert it to a third-person point of view." The Office says that where human-authored inputs are reflected in the output, they "contribute more than just an intellectual conception." Your original expression that is still perceptible remains protected. What the system added on its own is not.
See every path
- Did you write the words in this passage yourself (even if AI helped you brainstorm or outline first)? Yes → Did an AI tool then rewrite this passage, beyond minor fixes? Yes → Your original expression that survives in the output is still yours. The AI's own changes are not. If they are more than trivial, describe and exclude them when you register.
- Did you write the words in this passage yourself (even if AI helped you brainstorm or outline first)? Yes → Did an AI tool then rewrite this passage, beyond minor fixes? No → Protected as your work. The Copyright Office says assistive use, including brainstorming and outlining you only reference, does not affect protection.
- Did you write the words in this passage yourself (even if AI helped you brainstorm or outline first)? No → Did you substantially rewrite the AI text, or creatively select and arrange AI material inside your own writing? Yes → Your modifications and your selection and arrangement can be protected. The AI-generated text underneath is not, and must be excluded on a registration if more than de minimis.
- Did you write the words in this passage yourself (even if AI helped you brainstorm or outline first)? No → Did you substantially rewrite the AI text, or creatively select and arrange AI material inside your own writing? No → Not protected. Prompts alone, however detailed, don't make you the author of the output under current Copyright Office policy.
The cases: Zarya of the Dawn and Thaler
Two decisions anchor the rules, one from the Office and one from a federal appeals court.
Zarya of the Dawn (February 2023)
Kristina Kashtanova applied on September 15, 2022 to register Zarya of the Dawn, an 18-page comic book including its cover, and was granted registration without disclosing that its images came from Midjourney. In a February 21, 2023 letter, the Office concluded that she "is the author of the Work's text as well as the selection, coordination, and arrangement of the Work's written and visual elements," and that this authorship is protected. The Midjourney images were "not the product of human authorship." The Office said it would cancel her registration and issue a narrower one covering only the material she created.
Two details matter for novelists. Her text, which she wrote herself, was fully protected despite the AI images throughout the book. And small edits did not rescue the images: changes she made to a character's lips were "too minor and imperceptible to supply the necessary creativity for copyright protection."
Thaler v. Perlmutter (March 2025)
Stephen Thaler applied to register an image his AI system, the "Creativity Machine," produced, listing the machine as the sole author. The Office refused, and on March 18, 2025, the D.C. Circuit affirmed, holding that "the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being."
The opinion is careful about its limits. It says "the human authorship requirement does not prohibit copyrighting work that was made by or with the assistance of artificial intelligence," and that disputes over "how much artificial intelligence contributed to a particular human author's work" were not before it. Thaler asked the Supreme Court to take the case; the Court denied the petition on March 2, 2026. The human authorship rule stands.
The case still open: Allen v. Perlmutter
The harder question, how much human control over AI output is enough, is being litigated in Colorado. Jason Allen is challenging the refusal to register a Midjourney image. In its corrected brief filed February 6, 2026, the government notes that Allen says he entered prompts "at least 624 times," and argues that Midjourney, not Allen, determined the expressive elements. We found no decision in the case as of November 8, 2026: on the CourtListener docket, the last entries are June 2026 orders, and briefing closed in February 2026. If you see headlines about it, the ruling will matter for prompt-heavy work, not for novelists who write their own sentences.
- Sep 2022Zarya of the Dawn registeredA comic book made with Midjourney images was registered (effective September 15, 2022) without any AI disclosure. Source
- Feb 2023Zarya registration narrowedThe Office kept protection for the author's text and her selection and arrangement, but not the Midjourney images. Source
- Mar 2023Registration guidance issuedApplicants must disclose AI-generated content that is more than de minimis and describe the human contribution. Source
- Jan 2025Part 2 report on copyrightabilityAssistive AI use doesn't affect protection; prompts alone are not enough; no case made for extra protection of AI output. Source
- Mar 2025Thaler v. Perlmutter decidedThe D.C. Circuit held the Copyright Act requires a human author. Source
- Feb 2026Allen v. Perlmutter briefingThe government's corrected cross-motion for summary judgment was filed in a case over a Midjourney image its maker says took at least 624 prompts. Source
- Mar 2026Supreme Court declines ThalerCertiorari denied on March 2, 2026, leaving the D.C. Circuit ruling in place. Source
Copyright and KDP disclosure are different tests
It is tempting to assume that if Amazon calls something AI-assisted, it is safely theirs for copyright, or the reverse. The two systems ask different questions.
Amazon's content guidelines define AI-generated content by where it started. If an AI tool created the actual text, "it is considered 'AI-generated,' even if you applied substantial edits afterwards." Content you created yourself and then improved with AI tools, or text you wrote after brainstorming with AI, is "AI-assisted" and needs no disclosure.
The Copyright Office looks at what a human contributed to the final expression. Heavily rewritten AI text can contain protectable human modifications even though Amazon still counts it as AI-generated. Our guide to AI-assisted vs AI-generated writing covers the Amazon side in depth, and the KDP AI disclosure guide walks through the form.
A practical note on our own product, since it sits on this line. BlurbBio's AI analyzes and checks a manuscript rather than drafting it, but its paid rewrite presets can rewrite a passage you select. Treat that output as AI-generated text for Amazon, and for copyright assume only your own changes on top of it are yours.
How to register a novel that used AI
The March 2023 registration guidance says applicants "have a duty to disclose the inclusion of AI-generated content in a work submitted for registration" and to briefly explain what the human author contributed. In practice:
- Use the Standard Application. The guidance requires it for works with AI-generated material.
- Describe your authorship. In the "Author Created" field, describe what a human contributed. An applicant "who incorporates AI-generated text into a larger textual work should claim the portions of the textual work that is human-authored."
- Exclude the AI material. "AI-generated content that is more than de minimis should be explicitly excluded," under "Limitation of the Claim," in the "Other" field under "Material Excluded," with a description such as "[description of content] generated by artificial intelligence."
- Don't list the AI as an author. The guidance says applicants should not list an AI technology or its company as an author or co-author "simply because they used it."
- When unsure, say so. You may "simply provide a general statement that a work contains AI-generated material," and the Office will follow up. In some cases it will "explain that nothing needs to be disclaimed."
If you already registered a book without mentioning AI-generated material, the guidance says to contact the Public Information Office for a pending application, or file a supplementary registration for an issued one. Skipping that has a cost: applicants who don't update the record "risk losing the benefits of the registration."
Nothing in the guidance asks you to report AI you used only for brainstorming or for feedback on your own draft. Those uses don't put AI-generated material into the deposited work.
What to do on Monday
These are habits, not legal requirements, and they cost almost nothing.
- Keep your drafts and their dates. Version history is the simplest evidence of what you wrote and when. If you ever need to show your process, it is there.
- Log any AI-generated text you keep. A one-line note per chapter ("Ch. 12: two paragraphs of description generated, then rewritten") makes both a KDP disclosure and a registration easy.
- Use AI for feedback, not sentences, if ownership matters to you. A tool that flags a plot hole or a continuity slip leaves every word yours. A tool that writes the fix does not.
- Rewrite in your own words rather than editing machine prose. If an AI suggestion sparks an idea, close it and write the passage fresh. That is assistance; polishing its sentences is modification of AI material.
- Read your contract. If you are traditionally published or selling rights, check what your agreement says about AI use and warranties of original authorship before you use AI on the manuscript.
A brief note on the UK and EU
Outside the US the picture differs, and this post does not cover it in depth. The UK has protected some computer-generated works since 1988. Its March 18, 2026 report on copyright and AI notes that this protection "lasts for 50 years from the date of creation" and proposes that "in the absence of evidence of its ongoing value," it "should be removed, while copyright should continue to protect works created with AI assistance." That is a proposal, not a change in the law yet.
For the EU, the Copyright Office's report summarizes a 2024 policy questionnaire in which member states shared the view that AI-generated content may be eligible for protection "only if the human input in [the] creative process was significant," and that purely AI-generated works cannot be protected.
The one-sentence version
Write your own sentences and you own them; let a machine write them and nobody does, so if you keep any, disclose them to Amazon, exclude them from a registration, and keep notes on what you did. For the wider legal and market context, see our AI in publishing 2026 review, and if you are organizing a long manuscript, the free Story Bible tool and our guide to building one keep the canon in your hands.