Book Publishing

Amazon KDP News 2026: Your Author Survival Guide

KDP's pages on AI disclosure, accessibility and categories each ask for clearer information about a book: how it was made, who can use it and where it belongs.

A laptop showing a dark dashboard with charts, a blue banner reading KDP Survival Guide, a coffee cup and a pen on a desk.

Read one at a time, KDP's AI disclosure rule, its ebook accessibility guidelines and its category help page look separate. They share a theme: clear information about how a book was made, how well a reader can use it, and where it fits on the shelf.

This is our reading of KDP's pages, not a KDP statement. Each section links the page it relies on, so you can check what KDP itself says.

The practical takeaway: check your AI disclosure, your ebook file and your categories before you publish or update a book.

Three KDP Pages Worth Reading Together

KDP publishes separate help pages for AI-generated content, ebook accessibility and categories. They are linked in the sections below.

Why these topics belong in the same conversation

Authors often treat the AI rules, the accessibility guidelines and the category help page as separate items. They fit together as steps in one pre-publish check.

KDP's AI rule asks you to say whether AI generated content in the book. Its accessibility guidelines say how to prepare an ebook so more readers can use it. Its category help page says how many categories you can choose and what to do when a category is missing.

Practical advice: keep a note of how each book was made, check the ebook file, and pick categories that fit the book.

Habits Worth Checking

These habits are worth checking before you publish:

  • Treating upload as the end of the job: After publication, check that the listing, categories and disclosure still match the book.
  • Using vague production workflows: If you cannot say what AI helped create, what a human revised and what was licensed or adapted, you cannot answer KDP's disclosure question accurately.
  • Leaving technical quality for later: KDP's accessibility guidelines ask for text alternatives on informative images and for table captions and headings. Missing image descriptions and messy ebook formatting leave some readers without that information.
  • Chasing shelf space instead of fit: KDP lets you choose 3 categories per book (KDP Categories, read November 20, 2026). Choose categories that fit the book, not as many as possible.

The New KDP Landscape: A Strategic Overview

One way to read KDP's rules is as a single direction: books that are easier to trust, easier to use and easier to classify.

Fixing one item and skipping the others leaves gaps: a correct AI disclosure on an ebook with no image descriptions, or a careful category choice on a listing that does not match the book.

Habits That Fit KDP's Pages

In practice:

  • Clear provenance: If AI touched the manuscript or images, disclose it accurately.
  • Usable files: KDP's accessibility guidelines cover image descriptions, headings and tables. Clear headings and navigation are good practice for any ebook.
  • Metadata restraint: Better to fit a category well than to scatter a book into weak placements.
  • Reader-first packaging: Covers, descriptions and interiors should match each other, so readers get what the listing promises.

Treat disclosure and file quality as part of how the book is presented to readers, not as a separate chore.

Why this matters for long-term authors

If you publish one title and walk away, you may only notice a problem when something breaks. If you’re building a catalog, these checks apply to every launch.

Three Habits

Three habits cover most of this: document your workflow, format professionally, and choose categories with discipline.

Decoding KDP's AI Content Rule

Many authors think they must disclose any use of AI. KDP's rule is narrower.

KDP requires you to tell it about AI-generated content (text, images or translations) when you publish a new book or republish an edited one. Content you created yourself, where AI tools only helped you edit, refine or check it, is AI-assisted and does not need to be disclosed (KDP content guidelines: AI-generated content, read November 20, 2026). KDP says it will reject or remove content that does not follow its guidelines.

What KDP Asks

The practical distinction is simple. KDP's rule turns on whether AI generated the content, not on whether you used software anywhere in the workflow.

If an author uses AI to brainstorm titles, tighten a blurb, or suggest line edits, that is a different case from uploading chapters, illustrations, or translations generated largely by AI. A safe rule: if AI created material that appears in the final book, disclose it. KDP counts AI-generated content as AI-generated even if you edited it afterwards (KDP content guidelines: AI-generated content, read November 20, 2026).

What you should do before upload

Use a pre-upload checklist instead of relying on memory.

  1. Review your manuscript history: Identify whether AI drafted text, generated images, or produced translation output that remained in the final file.
  2. Label the workflow accurately: Distinguish between generation and assistance before entering the dashboard.
  3. Keep revision notes: If Amazon ever questions the book, notes help you show human oversight.
  4. Check your final files: Make sure the uploaded version matches what you intend to disclose.

Practical advice: If you’d be uncomfortable explaining your production process to KDP support in one paragraph, your workflow is too loose.

What you should not do

Avoid these:

  • Don’t under-disclose: Authors often try to classify generated material as “editing” because it sounds safer. If the tool produced final content, that’s the wrong move.
  • Don’t outsource blindly: If a freelancer used AI and didn’t tell you, you still answer KDP's disclosure question when you publish.
  • Don’t publish filler: Thin, repetitive, generic material serves readers poorly, whatever tools made it.

The policy gets easier once you standardize your process

A repeatable workflow is easier to answer from than ad hoc judgment. Build one intake form for every project. Ask who wrote the draft, who edited it, whether images were generated, and whether translation software played a direct role in the final manuscript.

The upside most authors miss

Disclosure sounds like a defensive task. Done accurately, it also keeps your own records clear, which helps if you republish or hand the book to someone else.

That’s the practical takeaway. Use AI if it helps your process. Keep human control. Disclose generated content accurately. Publish only what can survive scrutiny.

Mastering Accessibility for Broader Reach

KDP calls its ebook accessibility pages guidelines. In title setup, the default accessibility answer is "I don't know", and KDP says you can proceed with publishing with that selection (KDP Accessibility for eBooks, read November 20, 2026).

KDP's accessibility guidelines for ebooks say every image that adds information or context needs a text alternative of at most 140 characters, decorative images get a null alt attribute, and tables need captions and column or row headings. Without a text alternative, customers with low or no vision lose access to that information (KDP Accessibility for eBooks, read November 20, 2026). KDP also lets you describe your ebook's accessibility features during title setup.

Why Accessibility Helps Readers

Authors often hear “alt text” and think of legal compliance. KDP frames it around readers: without a text alternative, customers with low or no vision lose access to that information.

A well-structured ebook helps screen readers parse headings, links, lists, and images correctly.

What to fix in the manuscript file

Start with the elements most authors skip:

  • Image descriptions: Every meaningful image needs alt text that describes function or content, not decorative fluff.
  • Heading structure: Use real H1 through H6 hierarchy in EPUB production, not oversized bold text pretending to be headings.
  • Descriptive hyperlinks: “Click here” is weak. Link text should tell the reader what the destination is.
  • Clean navigation: A usable table of contents and predictable chapter structure help both readers and devices.

If you’re formatting in-house, a practical place to start is this guide on how to create an EPUB file. The goal is not just export. The goal is structured export.

A simple working method

You don’t need enterprise software to make major improvements. You do need discipline.

For Vellum users

Vellum simplifies clean chapter structure, but you still need to review image handling and exported navigation carefully. Don’t assume the software solved every accessibility issue automatically.

For Kindle Create users

Inspect the output manually on multiple devices or apps. Automated validation catches some problems, not all of them.

For InDesign users

Pay attention to tagged export choices, paragraph styles, and image metadata. InDesign can produce excellent EPUBs, but only if the underlying document is structured correctly. For EPUB files, KDP recommends the ACE accessibility checker from the DAISY Consortium (KDP QA Standards, read November 20, 2026).

A primary reference is the W3C's WCAG overview (read November 20, 2026). Use it as a sanity check before upload, especially if your book includes diagrams, screenshots, teaching visuals, or linked resources.

Which Books to Check First

Not every backlist title needs the same attention. Start with:

  • Illustrated nonfiction
  • Children’s books
  • Workbooks and guides
  • Any ebook with charts, screenshots, or instructional visuals

Text-heavy novels with minimal images usually need less remediation, though heading structure and navigation still matter. Books with complex visuals need a deeper pass.

Accessibility isn’t a niche feature added for a small audience. KDP lists it in its ebook guidelines, and it improves the file for every reader.

More categories sound helpful, but they do not guarantee more visibility.

That’s the problem behind one of the less discussed parts of KDP. You can choose 3 categories per book, and when Amazon adds a category that KDP does not list yet, KDP says to wait for its options to update and pick the closest match in the meantime (KDP Categories, read November 20, 2026). A very narrow category may have few shoppers, so check whether readers actually browse there.

A Caution on Narrow Categories

A common tip is to find narrow categories with lighter competition. That sounds logical, but a very narrow category may not have enough shoppers to help you. You can rank well inside it and still sell little if few readers browse there.

That is the trap with tiny categories. They create the appearance of precision without delivering market visibility.

A category is only useful if readers shop there. Low competition by itself isn’t a win.

How to Choose Categories

Think in terms of relevance first, traffic second, novelty last.

A practical method looks like this:

  • Start with reader intent: Where would a real buyer expect to find this book?
  • Check shelf strength: If a category is split into tiny fragments, be skeptical.
  • Avoid vanity placement: Ranking in a dead corner of the store doesn’t help your career.
  • Support with keywords: Strong categories and weak keywords still underperform. Weak categories and strong keywords also underperform.

For authors who want a clearer metadata foundation, this guide to tags for books is useful because it pushes you to think beyond category labels and into discoverability language readers use.

What works better than category chasing

A sound KDP metadata strategy uses three filters.

Established category with obvious buyer behavior

Choose categories with a recognizable browsing audience. Broad enough to have activity. Specific enough to fit the book properly.

Keyword alignment with the book’s promise

Your subtitle, description, backend keywords, and category choices should point to the same reader outcome. Scattered metadata creates confusion.

Positioning by sub-niche, not gimmick

A book aimed at a specific sub-niche can fit its readers better than a generic title on a crowded shelf. That works best when the positioning is consistent across title, cover, blurb, and keywords.

What to Avoid in 2026

These errors are easy to avoid:

MistakeWhy it hurts
Stuffing metadata with adjacent termsRelevance drops when signals become noisy
Using categories the book barely fitsReader dissatisfaction rises and conversion weakens
Never revisiting old listingsCategory maps change, and stale metadata drifts

The right mindset is less “How many slots can I occupy?” and more “Where will the right buyer find and trust this book?” That question leads to better category choices, better keyword choices, and cleaner long-term positioning.

Your Prioritized KDP Action Plan for 2026

Most authors don’t need more KDP commentary. They need order.

A workable response is a triage plan. Don’t update everything at once. Don’t panic-republish your entire catalog. Fix the highest-risk issues first, then improve discoverability where the upside is strongest.

The practical takeaway is simple: fix compliance items first, then work on positioning.

For new manuscripts

A new book should go through a pre-launch gate before you ever hit publish.

Stage one: compliance check

Handle the essential items first.

  • AI review: Identify whether any final text, images, or translations were AI-generated and prepare accurate disclosure.
  • Accessibility pass: Confirm image alt text, heading structure, navigation, and link clarity.
  • Format validation: Open the final file on actual Kindle apps or devices, not just your formatter’s export window.

If a manuscript fails any of these, stop there. Don’t move on to categories or ads yet.

Stage two: positioning check

After compliance, test the commercial fit.

  1. Clarify the reader promise: The cover, subtitle, and description should point to the same audience.
  2. Choose the sub-niche carefully: Look for focused demand, not broad ambition.
  3. Select categories conservatively: Avoid weak novelty categories unless you know they have active browsing.

Stage three: launch readiness

At this point, many authors still rush.

  • Description quality: Remove vague language and generic benefits.
  • Sample quality: The opening pages need to prove the book belongs on the platform.
  • Author workflow record: Keep internal notes on production choices, especially where AI or translation tools were involved.

For your existing backlist

Your backlist needs a different method. Don’t start with oldest first. Start with most exposed first.

Priority group one

Update books that are most likely to carry compliance risk:

  • heavily illustrated ebooks
  • books with charts, screenshots, or diagrams
  • translated editions
  • titles produced with mixed human and AI workflows

These books are the most likely to have accessibility or disclosure gaps.

Priority group two

Next, review books with sales history but weak current momentum. Those are often the best candidates for metadata cleanup and category correction because they’ve already shown some market signal.

Priority group three

Then assess books that were positioned too broadly. If the content is solid but the targeting is generic, reposition around a more specific audience or use case.

A sensible weekly workflow

Authors get overwhelmed because they think in “catalog overhaul” mode. Work in cycles instead.

Week focusPrimary taskSecondary task
Compliance weekAI disclosure auditAccessibility spot checks
Metadata weekCategory reviewKeyword refinement
Product weekCover and description alignmentSample-page quality review
Expansion weekSub-niche targetingEvaluate non-Amazon distribution

That last line matters. If your audience isn’t purely Kindle-centric, diversification can reduce dependence on a single marketplace. Amazon is still central, but it doesn’t have to be your only path.

The practical standard for 2026

A well-prepared KDP book needs four things working together:

  • Honest production disclosure
  • Readable, accessible file structure
  • Disciplined metadata
  • Clear sub-niche positioning

Miss one and the book can still publish. Miss several and performance gets harder to diagnose because the problems stack.

What Works and What Doesn’t

Here’s the blunt version.

What works

  • publishing fewer, better-prepared books
  • using AI as an assistant with human editorial control
  • choosing categories based on buyer behavior
  • upgrading proven backlist titles before chasing new experiments

What doesn’t

  • volume without oversight
  • pretending generated content is purely human-made
  • assuming formatting details are minor
  • spreading a book across weak discoverability lanes just to feel more visible

Conclusion: Prepare the Book Well

The useful part of KDP's pages is the pattern across them.

AI disclosure speaks to trust. Accessibility speaks to usability. Category discipline speaks to relevance. Together they describe a well-prepared book.

You do not need to fix everything at once. You do need a publishing process you can explain: how the book was made, whether the file works for readers, and why it sits in its categories.

Questions

Frequently Asked Questions About KDP

No. Start with your highest-risk titles first. Prioritize illustrated ebooks, books with complex formatting, translations, and any title created through mixed human and AI workflows. After that, move to books that already have some sales traction and could benefit from metadata or accessibility improvements.
Yes, but use it carefully. The issue isn’t tool usage by itself. The issue is whether AI-generated material appears in the final published work and whether you disclosed that accurately. Keep records of how you used the tools and make sure the final manuscript has real human oversight.
Start with image alt text, heading hierarchy, table of contents structure, and descriptive hyperlink text. Those changes usually produce the fastest improvement in file usability and are easier to audit than deeper layout refinements.
Not automatically. Choose categories based on fit and buyer behavior, not novelty.
It can. KDP says it will reject or remove content that does not follow its guidelines, and the AI disclosure question is part of them. When you’re unsure, review your workflow carefully before upload and err on the side of accuracy rather than minimal disclosure.
For EPUB files, KDP recommends the ACE accessibility checker from the DAISY Consortium, and screen-reader testing on common devices can help. Manual review still matters. Automated tools catch some issues, but they don’t replace reading through the finished ebook like a customer would.

Ready to publish your book?

Talk to a real publisher first. You leave the call with your exact price for your book.