KDP Compliance
KDP AI Disclosure Requirements: The Complete Author’s Guide
What Amazon requires you to disclose, what falls outside the policy, and the practical steps to stay compliant without second-guessing every AI tool in your workflow.
Amazon KDP introduced AI content disclosure requirements in 2023, and by 2026 the policy has teeth — titles removed, accounts warned, and enforcement tools quietly upgraded. Most KDP authors are now somewhere on the AI spectrum, from spell-checker to full manuscript generation, and the line between “disclosure required” and “disclosure not required” is where confusion — and real risk — lives.
This guide breaks down exactly what KDP requires you to disclose, what falls outside the policy, and the practical steps to stay compliant without second-guessing every AI tool in your workflow.
What Amazon KDP’s AI Disclosure Policy Actually Requires
Amazon added an AI content disclosure step to the KDP publishing workflow in late 2023. When you set up a new title — or update an existing one — you now see a section titled AI-Generated Content in the manuscript and cover upload stages.
The policy distinguishes two scenarios that require disclosure:
- AI-generated text: The body of the book, chapter text, or substantial portions were produced by a generative AI model (ChatGPT, Claude, Gemini, etc.) with minimal human rewriting.
- AI-generated images: Interior illustrations, book cover art, or other visual elements were created by an AI image generator (Midjourney, DALL-E, Stable Diffusion, Adobe Firefly, etc.) with minimal human editing.
- AI-generated translations: A translated version of the book was produced primarily by machine translation without substantial human editing by a qualified translator.
Note what the policy does not require disclosure for: using AI tools to improve, suggest, or refine work that you substantially created. This is the AI-assisted category, and it falls outside the disclosure checkbox.
The Disclosure Checkbox in Practice
When publishing or updating a title, KDP presents three separate disclosure checkboxes — one each for text, images, and translation. You check the boxes that apply. Leaving all boxes unchecked means you are declaring that no AI-generated content is present in those categories.
The disclosure is captured in your KDP account metadata. As of mid-2026, it does not appear as a visible label on the Amazon product page, though internal teams use it for review and enforcement decisions.
Where This Sits in the Publishing Workflow
The AI disclosure step appears during Book Details setup — specifically in the section after you upload your manuscript and cover. For existing titles, you access it through Bookshelf → Edit Book Details → scroll to the AI-generated content section. Updating the disclosure on a live title does not remove it from sale or trigger an automatic review.
| Content Type | Disclosure Required? | Examples |
|---|---|---|
| AI-generated text | Yes | Book written by ChatGPT with light editing |
| AI-generated images | Yes | Midjourney cover art, DALL-E interior illustrations |
| AI-generated translation | Yes | DeepL translation of full manuscript, unedited |
| Grammar/spell check (Grammarly, ProWritingAid) | No | AI-powered suggestions accepted/rejected by author |
| AI-assisted writing (substantial human rewrite) | No | AI outline, human-written chapters |
| AI background removal on author photo | No | Photoshop AI tools on a photo you took |
AI-Generated vs. AI-Assisted: The Line That Matters
The disclosure policy hinges on a distinction that Amazon defines but does not quantify with a percentage: the difference between AI-generated and AI-assisted content. Getting this right is the central compliance challenge for any author using modern writing tools.
What “AI-Generated” Means in Practice
AI-generated content is work where the generative AI model produced the substance of what the reader experiences. The clearest examples: a non-fiction book where you entered prompts into ChatGPT and published the output with minimal rewriting; a children’s picture book where every illustration was created by Midjourney with a text prompt; a translated edition produced by running the original through DeepL and uploading the result.
The word “minimal” is doing significant work here. Amazon has not defined a threshold. The working standard in the KDP author community — based on enforcement patterns and guidance from Amazon’s author forums — is that if the AI wrote the core creative content and you primarily edited for errors rather than rewriting for voice, it is AI-generated.
What “AI-Assisted” Means in Practice
AI-assisted work is where a human author created the primary content and used AI tools as aids. Using ChatGPT to generate chapter outlines that you then wrote from scratch: AI-assisted. Using Grammarly’s AI suggestions to catch passive voice: AI-assisted. Using Claude to brainstorm book titles that you then evaluated and chose from: AI-assisted. Using an AI image generator to create three rough concept sketches that you then hired a designer to redraw in a custom style: AI-assisted.
The test question: Would the work be substantially different if you had not used AI? If the AI produced the core text, images, or translation that ended up in the final book, it is AI-generated. If the AI saved you time but your creative decisions drove the output, it is AI-assisted.
The Grey Zone: Heavy AI Drafts Rewritten by Humans
The genuinely ambiguous case is where an author uses AI to generate a first draft of each chapter, then rewrites each chapter significantly — adding personal anecdotes, correcting facts, changing the voice, restructuring arguments. Where does that sit?
There is no official Amazon guidance for this scenario. The practical approach most compliant authors use: if a reader who knows your writing would recognize the final work as yours — your voice, your examples, your structure — it is AI-assisted. If they would say “this reads like it was written by a chatbot,” disclose it. When in doubt, disclosing is always lower risk than not disclosing. You can always update the disclosure later.
What Happens If You Don’t Disclose
Amazon enforces AI disclosure through a combination of automated detection and manual review. Both have escalated since 2023. The consequences of undisclosed AI content are not theoretical — they have affected thousands of titles.
Title Removal
The most common enforcement outcome is title removal. Amazon sends an email notifying you that your title has been removed for violating content guidelines, typically citing the AI disclosure policy or a related quality standard. The title is removed from the store immediately. Any pending royalty payments for the review period are typically withheld pending the review resolution.
Account Warning and Suspension
Repeated violations — or a single egregious one — can result in a publishing privileges warning. Amazon issues a formal notice that your account is under review. In severe cases, particularly for authors who published large volumes of AI-generated content without disclosure, account termination has occurred. A terminated KDP account loses access to the dashboard, all published titles, and any unpaid royalties above the payment threshold.
The Detection Reality
Amazon uses AI detection tools internally. These tools have improved significantly and are applied at the review stage for new titles and in batch audits of existing catalogues. False positive rates remain a real issue — human-written content in certain styles (technical, instructional, formal business writing) can score high on AI detectors. But the enforcement trigger is not a detection score alone; Amazon combines detection signals with quality reviews and account behaviour patterns.
How to Disclose AI Content Correctly: Step-by-Step
Correct disclosure is a two-minute process during title setup. Here is the exact workflow for both new titles and updates to existing books.
For a New Title
During the KDP publishing process, after uploading your manuscript and cover, you reach the Book Details page. Scroll to the section labelled AI-Generated Content. You will see three questions:
- Does this book contain AI-generated text? (Yes / No)
- Does this book contain AI-generated images? (Yes / No)
- Does this book contain AI-generated translations? (Yes / No)
Select the appropriate answers and continue. You cannot skip this step — the form requires a response to proceed to the pricing and rights section. Once the book publishes, the disclosure is recorded in your account metadata.
For an Existing Published Title
Log in to KDP. On your Bookshelf, click the three dots next to the title and choose Edit Book Details. Navigate to the AI-generated content section (it appears in the same location as during initial setup). Update the disclosures as needed and click Save and Continue. The changes take effect within 24–48 hours. The book does not go into review for this change alone.
Auditing Your Back Catalogue
If you published titles before the disclosure requirement was introduced (pre-2024) or during early 2024 when many authors missed the new fields, it is worth auditing your catalogue. Export your bookshelf data from KDP Reports, then work through each title and check whether any contain AI-generated elements that were not disclosed. Update any that require it. This proactive audit significantly reduces the risk of a batch enforcement action.
AI Tools That Don’t Trigger Disclosure
Not every AI tool creates a disclosure obligation. Understanding which tools fall outside the policy helps you use your existing workflow without unnecessary compliance friction.
Writing and Editing Tools
Grammarly, ProWritingAid, and Hemingway Editor — even with their AI-powered suggestion features — are writing aids. They do not generate content; they suggest improvements to content you wrote. No disclosure required. The same applies to Microsoft Word’s AI editing suggestions, Google Docs Smart Compose, and similar inline assistants.
Using ChatGPT to check grammar or improve sentence clarity — where you review and accept or reject each suggestion — sits in the AI-assisted zone. If you paste a paragraph into ChatGPT and publish the rewritten output verbatim, that paragraph is AI-generated. If you review the suggestion and rewrite in your own words, it is AI-assisted. The key is whether your creative judgment shaped the final sentence.
Research and Planning Tools
Using AI to research topics, generate reading lists, summarise long documents for your notes, or outline a chapter structure you then write from scratch: none of these require disclosure. The AI produced inputs to your thinking, not content in your book. Tools like Kip Scout for market research or Bookbeam for keyword analysis use AI in their analysis engines but don’t generate your book content. For a detailed comparison of keyword research tools, see our KDP Spy vs Publisher Rocket review.
Cover Design Tools with AI Features
Canva, BookBrush, and similar cover design platforms have added AI-powered features — background removal, object erasure, style suggestions. Using these tools on photos you took or licensed images does not make the cover AI-generated. The AI is assisting your design process. If you used a text prompt to generate the cover image from scratch in an AI generator, that is AI-generated and requires disclosure.
AI Disclosure and KDP Select: What Changes
KDP Select enrollment adds an additional layer of Amazon scrutiny to titles. Select titles must be exclusive to Amazon, must meet quality standards, and are reviewed more carefully for compliance because they participate in Kindle Unlimited and the promotional programme. The AI disclosure policy applies identically — there is no Select-specific exemption or stricter standard beyond the base policy.
Quality Standards in KDP Select
KDP Select has historically been stricter about content quality than standard KDP publishing. AI-generated titles that pass initial publishing review have sometimes been removed from Kindle Unlimited specifically, even when the standard ebook remains available. Amazon does not publish separate quality metrics for Select, but the enforcement pattern suggests that AI-generated content in Select faces higher scrutiny than the same content in standard distribution.
If you are publishing AI-generated content and enrolling in Select, the combination of AI disclosure and Select enrollment places your title in a higher-review category. This is not a reason to avoid disclosing — the opposite: it is a reason to disclose accurately and ensure the title meets the quality bar that makes AI-generated content worth publishing at all. Understanding how to price your KDP titles correctly becomes more important when your title is under additional scrutiny.
KDP Select and AI Translation
The AI translation scenario is particularly relevant for Select because many authors run machine translation on English titles to publish in German, Spanish, French, and other Kindle markets. Select requires that translated editions be separate titles enrolled in their own markets. A machine-translated title without human editing and without disclosure is one of the highest-risk combinations in the 2026 enforcement environment.
The Business Case for Transparent Disclosure
Compliance aside, there is a practical business argument for accurate disclosure that goes beyond avoiding enforcement. The KDP market is becoming more sophisticated about AI content quality, and readers are more discerning than the 2023 flood of low-quality AI books suggested.
Reader Trust and Reviews
Kindle Unlimited readers who feel they received low-quality AI-generated content without indication tend to leave negative reviews and use the “report content” function. Understanding how KU page-reads contribute to royalties is covered in our Helium 10 Cerebro vs Publisher Rocket comparison. These signals feed into Amazon’s quality review queue. High return rates — another quality signal Amazon tracks — correlate with AI-generated content that doesn’t deliver on the book description’s promise.
Authors publishing disclosed AI-assisted content that is genuinely useful tend to see better review patterns because the disclosure sets appropriate expectations. Readers who buy a book knowing it was AI-assisted and find it valuable rate it positively. Readers who feel misled do not.
Long-Term Catalogue Safety
Building a KDP catalogue on undisclosed AI content is a compounding risk. As Amazon’s enforcement improves, older undisclosed titles become increasingly likely to be flagged. Authors who have 50 or 100 undisclosed AI titles face potential catalogue-wide enforcement action. Authors who disclosed from the start have no back-catalogue exposure. If you are building a KDP business for the long term, transparent disclosure from the first title is the only sustainable approach. This connects to the broader question of how to build a KDP business that lasts rather than chasing short-term catalogue volume.
2026 Policy Updates: What Changed and What’s Coming
Amazon has not published a formal changelog for its AI content policy, but enforcement patterns and KDP community reports document several shifts since the initial 2023 policy introduction.
What Changed in 2025–2026
| Period | Change | Impact |
|---|---|---|
| Late 2023 | AI disclosure checkboxes added to publishing workflow | Low enforcement, awareness-building phase |
| Mid 2024 | Batch removal of low-quality AI titles in certain categories | First significant enforcement wave; romance, self-help, children’s hardest hit |
| Early 2025 | AI detection integrated into initial review pipeline | New AI-generated titles caught before going live more frequently |
| Late 2025 | Account-level enforcement for repeat violations | Suspension and termination of high-volume undisclosed AI publisher accounts |
| 2026 | Enforcement stabilisation with clearer quality floor | Disclosed AI content meeting quality standards survives; thin undisclosed AI removed systematically |
What to Expect in Late 2026 and 2027
The direction of Amazon’s AI content policy is toward more transparency, not less. Industry analysts and KDP community observers anticipate several developments: public-facing AI disclosure labels on product pages (giving readers the choice to filter), stricter minimum quality standards for AI-generated titles, and possible royalty differentiation for AI-generated versus human-authored content in Kindle Unlimited.
None of these are confirmed Amazon plans, but the trajectory is consistent with how Amazon has handled other content quality initiatives. Authors who have built their catalogue on transparent AI use — disclosed, quality content — are well-positioned for any of these scenarios. Authors relying on volume of undisclosed AI content are not.
Practical Disclosure Strategy for KDP Authors in 2026
Given where enforcement sits in 2026, here is a practical framework for managing AI disclosure across your KDP catalogue.
The Three-Tier Audit
Tier 1 — New titles: Before publishing any new title, run through the three disclosure questions honestly. If any element is AI-generated, check the box. This takes two minutes and eliminates the primary compliance risk for new content. Note your disclosure decision in your title records so you have documentation if Amazon ever requests clarification.
Tier 2 — Post-2023 titles you haven’t audited: Go through titles published since late 2023 and check whether the AI disclosure fields are accurately filled. These are your highest-risk titles because the policy was active but many authors missed the new fields during initial setup. Update any that are inaccurate.
Tier 3 — Pre-2023 titles: These were published before the disclosure requirement existed. Amazon is not retroactively penalising pre-policy titles for missing historical disclosure. However, if you used AI-generated content in these titles and have not updated the disclosure fields, consider doing so as a precaution — particularly if those titles are in categories Amazon has been actively auditing (self-help, children’s, romance, business).
Documentation That Protects You
If Amazon ever flags one of your titles, documentation of your authorship process is your primary defence. For human-written content: keep draft versions, writing notes, and revision history. For AI-assisted content: document your process — the prompts you used, how you rewrote the output, what percentage of the final text is original. For correctly disclosed AI-generated content: the disclosure itself is your documentation.
Cloud storage of drafts (Google Drive, Dropbox) with timestamp history provides verifiable evidence of a genuine writing process. For authors who use web research tools in their writing process, our Firecrawl review covers how to use AI-assisted research without generating disclosable content. The same discipline that helps you understand the difference between publishing routes applies to understanding what documentation each route requires.
Frequently Asked Questions
Does Amazon KDP require AI disclosure in 2026?
Yes. Amazon KDP requires authors to disclose AI-generated content during the publishing process. The disclosure applies to AI-generated text, images, and translations. AI-assisted content — where a human substantially wrote or created the work and used AI only as a tool — does not require the AI-generated disclosure checkbox.
What is the difference between AI-generated and AI-assisted content on KDP?
AI-generated means the content was produced by an AI system with minimal human creative input. AI-assisted means a human author created the core work and used AI tools for editing, suggestions, or refinement. KDP requires disclosure only for AI-generated content.
What happens if I don’t disclose AI-generated content on KDP?
Consequences can include removal of the title, account warning, suspension of publishing privileges, or account termination. Amazon has invested in AI detection tools and has been removing titles that appear AI-generated without proper disclosure.
Does using Grammarly count as AI-generated content?
No. Grammar checkers and style tools are writing aids, not content generators. They do not trigger the KDP AI disclosure requirement.
Can I update my AI disclosure after a book is already published?
Yes. Update it through KDP Bookshelf → Edit Book Details → AI-generated content section. The book does not get unpublished or go into review for this change alone.
Are AI-generated book covers affected by the KDP disclosure policy?
Yes. If your cover was generated by an AI image tool (Midjourney, DALL-E, etc.) with minimal human editing, it requires disclosure. Using AI tools to touch up a photo you took is AI-assisted and does not require disclosure.
Does the AI disclosure policy apply to KDP Select books?
Yes. The policy applies to all KDP titles regardless of Select enrollment. AI-generated content in Select faces higher scrutiny but the same disclosure requirement applies.
Will Amazon remove my book if it detects undisclosed AI content?
Amazon has removed books flagged as undisclosed AI-generated content. Enforcement has increased since 2024. The safest approach is to disclose if in doubt — you can always update the disclosure, but you cannot un-remove a taken-down title without an appeals process.