KDP AI content disclosure for coloring books [2026]

Amazon KDP requires publishers to declare AI-generated text, images, or translations at upload [1]. For AI-generated coloring pages, this means checking the box: the line art was produced by an AI tool. Editing-only use (cleanup, line tweaks, contrast adjustment) counts as AI-assisted and is not declared [1]. The disclosure is an internal compliance record: no public-facing badge on the listing, no search-ranking penalty, no royalty change, no category restriction [5]. Non-disclosure risks title removal and, on repeat violations, account suspension [5][6].

TL;DR:

• KDP added the AI content disclosure question on September 7, 2023 [2]. It applies to every new title and every edit of an existing title.
• AI-generated means the AI tool produced the actual content readers see (text, image, or translation), even if you applied substantial edits afterwards [1]. For coloring books, this is almost always the line art.
• AI-assisted means you created the content and used AI only for edits, refinement, or error-checking [1]. Not declared.
• The disclosure is invisible to buyers. It does not affect ranking, royalties, ads, or category placement [5].
• Failing to disclose is the policy violation, not declaring. Honest disclosure clears the book; non-disclosure can pull it down.

If you are also auditing the file itself for rejection triggers (bleed, gutter, DPI, font embedding, trim, cover wrap math, content policy), the manuscript pre-flight checker walks every documented rejection layer including AI disclosure. This post covers the AI question in depth: what it covers, what it does not, and what changes after you check it.

Table of contents

• What does Amazon KDP's AI content disclosure question actually ask?
• What counts as AI-generated vs AI-assisted for coloring books?
• What happens after you declare AI use?
• What happens if you do not declare?
• Are AI-generated coloring pages copyrightable in the US?
• How should AI-generated coloring book publishers actually answer the question?
• Common edge cases

What does Amazon KDP's AI content disclosure question actually ask?

The 2026 form asks whether the title contains AI-generated content, with the disclosure text reading: "This title contains AI-generated content. This includes text (e.g., chapters, sections), images (e.g., illustrations, diagrams, cover art), or translations that were produced using AI tools" [5]. It is a single binary checkbox per format, presented inside the publishing flow alongside the other listing inputs, and it applies to every new title and every meaningful edit of an existing one [1][2].

Amazon introduced the question on September 7, 2023, after months of pressure from the Authors Guild and a wave of obvious-AI titles flooding the bestseller charts in early 2023 [2]. Before the change, KDP had no AI-specific question at all. Now the question sits as a required field, and skipping it (or worse, answering incorrectly) is a Terms of Service violation rather than a product-quality issue.

The question covers three categories of content, in this order in Amazon's own language [1]:

• Text. Chapters, sections, descriptions, summaries, or any other words readers see in the final book. Includes everything from a 5,000-word foreword draft to a 12-word activity prompt.
• Images. Illustrations, diagrams, cover art, decorative elements, line art for coloring pages, anything visual that appears in the printed product.
• Translations. Machine translations that form the published edition. A book originally written in English and published in Spanish via an AI translation tool is in scope.

For a coloring book made with an AI image generator, the relevant category is images. The text content (title page, copyright page, brief intro) is almost always human-written, so the text category typically does not apply. Translations only matter if you used an AI tool to localize the book.

Where the question appears in the publishing flow

The AI content question lives inside the "Book details" or "Content" step of the KDP publishing form, after the title and contributors but before the categories and keywords [5]. It does not appear on the cover-upload step or the pricing step. You see it once per format: if you publish both paperback and hardcover editions, each format gets its own disclosure. The Kindle edition (if any) gets a separate disclosure too.

What counts as AI-generated vs AI-assisted for coloring books?

For coloring books, the split is clean. AI-generated covers any page where an AI image generator produced the line art (the actual drawing the reader colors), even if you traced it, adjusted lines, removed artifacts, or upscaled it after generation [1]. AI-assisted covers human-drawn art that you ran through an AI tool only for refinement, like noise removal on a scanned hand drawing, contrast cleanup, or vectorization of pencil sketches [1][6].

The KDP help page makes the distinction explicit: "If you used an AI-based tool to create the actual content (whether text, images, or translations), it is considered 'AI-generated,' even if you applied substantial edits afterwards" [1]. Edits do not change the classification. Substantial edits do not change the classification. The classification turns on where the original content came from, not how much you reworked it.

Concrete examples for coloring book publishers

AI-generated, must disclose:

• Coloring pages generated by Midjourney, DALL-E, Stable Diffusion, or any other text-to-image model, even after extensive cleanup in Procreate or Photoshop [1].
• AI-generated cover art, even if you added title text and your name manually.
• AI-generated decorative elements on the title page or back cover.
• A coloring page where you wrote the prompt, the model rendered it, and you traced over the result. The original line work is AI-generated, so the page is AI-generated [4].

AI-assisted, do not disclose:

• Hand-drawn line art that you photographed and ran through an AI background-removal or noise-removal tool [6].
• Hand-drawn art that you vectorized with an AI-powered vectorizer (the underlying drawing is yours, the AI only converted format).
• AI-edited title page copy where you wrote the words and used a grammar checker or rephrasing tool.
• AI brainstormed coloring page topics where the actual drawings are hand-done.

Gray-area cases:

• A coloring page where AI generated a rough composition and you redrew it by hand from scratch. The final art is human-authored. The Authors Guild interpretation suggests this is human-authored and not disclosed [3], but conservative publishers disclose to stay clear of any "substantial edits afterwards" reading [1]. The US Copyright Office position (see further down) treats heavy human reworking of AI output as still requiring disclosure for copyright registration purposes [3][4].
• A coloring book that mixes 30 AI-generated pages and 30 hand-drawn pages. The book contains AI-generated content, so the box gets checked. There is no partial disclosure; the disclosure is binary [5].

The honest rule of thumb: if the AI model produced the line art that ends up in the printed book, declare. Edits, cleanup, and adjustments do not change the answer.

What happens after you declare AI use?

Nothing visible to your buyers. The disclosure is an internal compliance record. It does not appear as a badge or label on the customer-facing product page, does not affect search ranking, does not change ad eligibility, does not affect royalty rates, and does not restrict category placement [5]. Amazon uses the disclosure data internally for compliance review and may apply more careful review to the file [6], but none of that surfaces to the reader.

Specifically, the 2026 policy text confirms: "Disclosure does not affect search ranking, visibility, ad eligibility, or promotional program participation" [5]. KDP Select eligibility (Kindle-only) does not change. The standard distribution royalty (60% for the $2.99 to $9.99 band on paperback) does not change [5]. The 3-category limit and the 7 keyword slots do not change.

The one operational change worth knowing about: Amazon may flag the title for a longer or more careful manual review when AI disclosure is checked, particularly for visual-heavy books [6]. This shows up as a 24 to 72 hour longer review window before the book goes live. New publishers occasionally interpret this as a rejection signal; it is not. It is process, not penalty.

What the disclosure does change

The disclosure shifts compliance liability. Once you declare AI use, Amazon has on-record acknowledgement that the content came from an AI tool, which puts the burden of policy compliance squarely on you [1]. The KDP content guidelines confirm: publishers remain "responsible for ensuring that all AI-generated content adheres to our content guidelines and complies with intellectual property rights" [1]. Declaring does not shield you from trademark, copyright, or content policy violations downstream.

The disclosure is also permanent for that edition. Once recorded, it stays on the internal record [5]. If you re-upload a revised manuscript that swaps AI-generated pages for human-drawn ones, you cannot retroactively un-check the box on the original edition. The path is to publish a new edition with a new ASIN if the content origin changes meaningfully.

What happens if you do not declare?

Non-disclosure is the actual policy violation, not declaring. KDP treats failing to disclose AI-generated content as a Terms of Service breach, and the documented consequences scale from title removal on the offending book to full account suspension on repeat violations [5][6]. The Authors Guild's coverage of the original 2023 policy made the enforcement framing explicit: the disclosure is mandatory, and Amazon plans to act on non-compliance [2].

Industry coverage of 2024 and 2025 enforcement patterns suggests Amazon does not actively scan covers to detect AI generation (the technical capability exists but the cost of false positives is high). Detection comes from a few specific paths:

• Reader reports. A buyer who recognizes obvious AI-generation patterns (extra fingers, melted faces on character pages, characteristic AI artifacts in the line work) can report the listing through the standard customer feedback channels.
• Competitor reports. Other KDP publishers who recognize AI patterns may report listings, particularly when an undisclosed AI book ranks above their hand-drawn one.
• Pattern matching at scale. Amazon's catalog systems run periodic sweeps for catalog-level patterns (identical prompt-output signatures across many publishers, suspicious mass-upload patterns) and flag matching titles for review.
• Manual review escalation. Any review prompted for another reason (bleed mismatch, cover spec error, category dispute) can surface an undisclosed AI flag if the reviewer judges the content visibly AI-generated.

Once a title is flagged for undisclosed AI, the path is removal and a publisher warning. Repeat violations on the same KDP account trigger an account-level action, which can include publishing rights revocation [6]. The asymmetry matters: honest disclosure has zero downside (no public badge, no ranking penalty), and the only downside path is non-disclosure being caught.

For AI-generated coloring book publishers specifically, the practical risk concentrates around mass-uploaded catalogs (one publisher pushing 50 books in 3 months, each visibly AI-generated, none disclosed). Single-title publishers who disclose honestly do not face this risk pattern.

Are AI-generated coloring pages copyrightable in the US?

Not as standalone art [3][4]. The US Copyright Office published Part 2 of its AI report in January 2025 confirming that AI-generated material cannot receive copyright protection in the United States, because copyright protection requires human authorship and prompts alone do not constitute sufficient creative control [3][4]. This applies to wholly AI-generated content and to AI-generated elements within otherwise human-authored works.

The Office's Part 2 report was explicit on prompting: "Prompting alone, even using complex and multiple prompts, will not provide copyright protection in the output" [3][4]. The Authors Guild summary adopted the Office's framing that iterative prompting resembles chance rather than creative control, which means the AI tool, not the prompter, is doing the creative work for copyright purposes [3].

What this means for an AI-generated coloring book

The book as a whole can still be copyrighted, but the AI-generated images inside are filtered out for infringement analysis purposes [3]. The copyrightable parts of an AI-generated coloring book are:

• Selection and arrangement. The order you placed the pages, the themes you chose to combine, the front and back matter you wrote.
• Human-authored text. Title page, copyright page, intro, instructions, any human-written content.
• Cover layout and design (the human-authored composition, even when individual visual elements came from AI).
• Compilation choices. The 60 pages you selected from 200 AI generations.

The AI-rendered line art on each individual page is not copyrightable by you. If a competitor copies the exact same prompts to a different AI tool and generates similar pages, they have not infringed your copyright on the line work, because you do not hold copyright on the line work [3][4].

How this connects to KDP disclosure

The KDP disclosure question is separate from copyright registration. KDP disclosure is a contract requirement to Amazon under their Terms of Service [1]. Copyright registration is a legal action with the US Copyright Office [3]. The two do not share data, but both ask similar questions about what was AI-generated. If you register the copyright on a coloring book with AI-generated line art, you must disclaim the AI-generated portions to the Copyright Office [3], regardless of what you disclosed to KDP.

For practical purposes: declare to KDP, and if you also register with the Copyright Office, register only the human-authored elements (selection, arrangement, text, original cover composition) and disclaim the AI-generated visual elements [3].

How should AI-generated coloring book publishers actually answer the question?

Check the box if any AI tool produced any visual content that appears in the printed product. Do not overthink the percentage of pages, the amount of post-editing, or whether the AI tool was "just a starting point." The KDP guideline is unambiguous: any AI-generated content triggers the disclosure, and substantial edits do not change the classification [1]. The honest, conservative, lowest-risk answer is to disclose when in doubt.

The 5-minute decision tree:

1. Did an AI image model render any line art that appears in the printed pages? If yes, disclose. (Includes Midjourney, DALL-E, Stable Diffusion, Leonardo.AI, Recraft, Ideogram, Flux, and any other text-to-image tool.)
2. Did an AI text model write any words readers see? If yes, disclose. (Rare for coloring books, but applies to AI-written intros, activity prompts, or back-matter blurbs.)
3. Did you use AI only for editing or refinement on content you originally created? If yes, do not disclose. (Background removal, line cleanup, color correction, grammar checks, vectorization of your scans.)
4. Mixed authorship across pages? If any single page in the book is AI-generated, disclose [5].

The pre-flight pass on this question covers it in one check: did you correctly answer KDP's AI Content question for AI-generated images? The manuscript pre-flight checker flags this as critical because mis-declaring is one of the few content policy items that can trigger account-level consequences rather than just title-level ones.

Practical workflow notes

Industry guidance, including the Brandon Rohrbaugh KDP rules summary, recommends keeping a workflow log: which tool you used, when you generated each batch, what prompts you used, which pages came from which source, and what edits you applied [6]. The log is not required by KDP and is not requested at upload, but it produces a paper trail that helps if Amazon ever asks for clarification after a customer report or a competitor complaint.

The other industry-best practice: separate AI generations from human work in your asset directory. If you mix them, you lose track of which page is which, and the disclosure question becomes guesswork on a re-upload years later. AI-generated batches in one folder, hand-drawn art in another, edited variants of either in clearly labelled subfolders.

Common edge cases

Does tracing over AI output change the answer? No. The KDP guidance treats substantial edits, including tracing and redrawing, as not changing the AI-generated classification, because the original content came from the AI tool [1]. The Copyright Office takes the same position [3]. Tracing also does not produce a copyrightable derivative work, because the original is not yours to derive from.

What about AI upscalers and AI background removers on human-drawn art? AI-assisted, not declared [6]. The original art is yours; the AI tool is only performing a refinement task on existing content. This is exactly the case the KDP guideline carves out for non-disclosure.

Does declaring AI use affect KDP Select eligibility? No [5]. KDP Select is the Kindle-only exclusivity program and does not apply to paperback or hardcover coloring books anyway. Even for Kindle, the AI disclosure has no impact on Select enrollment or page-read royalties.

Does declaring AI use limit category placement? No [5]. The 3-category picks work the same way regardless of disclosure. You can lock the same picks with the category browser and the same backend keywords with the keyword optimizer whether or not you declared AI.

What if I declared AI when I should not have (false positive)? No mechanism to undeclare on the current edition [5]. The practical fix is to leave the disclosure as-is on that edition; the over-disclosure is harmless because the disclosure has no negative consequences. If accuracy on the record matters (e.g., for legal or trademark purposes), publish a new edition with corrected disclosure.

What if I did not declare and want to fix it on an existing title? Re-open the title in KDP, navigate to the AI content question, check the box, save. Amazon documents this path for correcting non-disclosure on already-published titles, and proactively correcting reduces enforcement risk relative to being caught by a customer or competitor report [6].

Frequently asked questions

Is AI-generated content allowed on Amazon KDP?

Yes, with disclosure. KDP does not ban AI-generated content; it requires that publishers declare it [1][2]. The policy framing is transparency, not prohibition. Coloring books generated entirely by AI are publishable so long as the AI Content question is answered honestly and the content otherwise complies with KDP policies on trademarks, copyright, and content quality.

Does the AI disclosure question apply to my book's existing ASIN or only new uploads?

Both. The September 2023 policy applies to every new title at first publication and every meaningful edit of an existing title [1][2]. If you re-uploaded a manuscript or cover after September 2023 and AI was used in the content, the disclosure should be on the current record. You can update it on existing titles through the Books > Bookshelf > Edit listing flow.

Will buyers see that my coloring book is AI-generated?

No. The disclosure does not appear on the product detail page, in search results, in customer recommendations, or anywhere visible to buyers as of 2026 [5]. It is an internal compliance record between you and Amazon. The Authors Guild has advocated for reader-facing disclosure but Amazon has not implemented it [2].

Can I copyright an AI-generated coloring book?

The book as a compilation can be copyrighted (selection, arrangement, human-authored text, original cover composition), but the AI-generated visual content inside cannot [3][4]. If you register with the US Copyright Office, disclaim the AI-generated portions in the application and limit your claim to the human-authored elements.

Does declaring AI use affect my book's reviews or visibility?

No. The disclosure has no impact on search ranking, ad placement, customer reviews, or any other public-facing signal [5]. Books with disclosure perform identically to books without in Amazon's surfacing algorithms.

What happens if Amazon detects undisclosed AI content?

Title removal on first offense, account-level action (including publishing rights revocation) on repeat offenses [5][6]. Detection comes from customer reports, competitor reports, manual reviews triggered for other reasons, and Amazon's periodic catalog-level pattern sweeps.

From disclosure to a clean upload

The AI content disclosure is one of 28 documented KDP rejection triggers across interior, cover, metadata, and content policy. Running the manuscript pre-flight checker walks every documented trigger in one pass, including the AI disclosure check and the trademark and derivative-content checks that often travel with it. For the format-side rejection triggers (bleed, gutter, DPI, trim, font embedding) that gate most pre-upload rejections, see KDP coloring book format guide and the bleed mistakes deep dive.

If you are earlier in the workflow and still picking AI prompts that produce KDP-quality line art, see best AI prompts for KDP coloring books for the patterns that match the bold-and-easy and detailed style standards Amazon buyers expect. The creation guide covers the full end-to-end production sequence from prompt to upload.

The disclosure question is short and the answer is usually obvious. The mistake to avoid is treating it as optional or strategically ambiguous. AI-generated coloring books publish without issue when the disclosure matches reality. The only downside path is non-disclosure being caught, which is asymmetric in the wrong direction for any publisher trying to build a sustainable KDP catalog.