TL;DR: Monthly new e-book releases on KDP nearly tripled between 2022 and 2025, with economists estimating roughly two-thirds are now AI-assisted. Amazon has escalated from removing individual books to account-level enforcement for undisclosed AI content. Authors and publishers who use AI tools and fail to disclose correctly now risk account suspension, not just book removal.
Amazon's self-publishing platform has always attracted high-volume publishers, but the scale of AI-generated content now flowing through Kindle Direct Publishing represents a different order of magnitude. The compliance picture has changed with it: what started as a disclosure checkbox has become the foundation for Amazon's enforcement architecture.
The scale of the surge
A 2025 NBER working paper by economists Imke Reimers and Joel Waldfogel found that the monthly rate of new e-book releases on KDP nearly tripled between 2022 and late 2025. Researchers estimate that roughly two-thirds of new books now involve AI assistance in generation or editing.
Amazon's own data shows over 40% of new KDP titles in Q1 2026 involved AI assistance in at least one stage of production, up from an estimated 15% in 2024.
The pattern within this surge matters for enforcement. New post-LLM entrants to the platform dominate the volume increase. Their books cluster at the bottom of sales rankings: poor engagement, low ratings, minimal reader traction. What observers call "AI book slop" tends to be generic content published at high velocity across categories, sometimes with thin reviews manufactured to build initial visibility.
The cumulative effect on search ranking is real even if each individual title sells almost nothing. Fifty AI-produced titles in a niche category create enough category noise to push legitimately authored books further down search results. This is why Amazon's enforcement response has targeted the underlying business model, not just individual titles.
Amazon's disclosure requirement
KDP requires authors to declare AI involvement during the publishing process. The requirement covers three categories:
- AI-generated text: content where the substantive prose was produced by an AI tool, even with significant human editing afterward
- AI-generated cover art: images produced by AI image generation tools
- AI-generated translations: machine translations produced using AI tools rather than human translators
Amazon's definition is intentional: if you used an AI-based tool to create the actual content, it is considered AI-generated regardless of how much editing you applied. The threshold is at creation, not at final output.
The disclosure is made during the KDP publishing workflow. It does not appear on the book's product page and is not visible to buyers. It is an internal compliance record.
For the practical mechanics of exactly what to check and when, the Amazon KDP AI disclosure compliance guide covers the checkbox workflow, the AI-assisted vs AI-generated distinction, and a decision flowchart for edge cases.
How enforcement has escalated
Amazon's enforcement posture shifted noticeably in 2025 and has continued through 2026. The platform now uses a combination of automated detection and human review to identify undisclosed AI content. Automated detection analyzes writing patterns, metadata consistency, and submission velocity. High-volume accounts publishing dozens of titles per month in unrelated categories attract elevated scrutiny.
The key enforcement escalation is the move from title-level to account-level action. Previously, a flagged book would be removed. Now, publishers with a pattern of undisclosed AI content or consistently low-quality AI books face account-wide publishing restrictions. Amazon has also begun reviewing previously published titles rather than limiting review to new submissions. Books published before the disclosure requirement was introduced that appear to be AI-generated have been flagged, with publishers asked to update their disclosure status retroactively.
Enforcement categories and the actions they trigger:
| Violation pattern | Amazon action |
|---|---|
| Single undisclosed AI title, first offense | Book removal, disclosure update required |
| Repeated undisclosed AI content | Account-level publishing restrictions |
| Egregious or systematic non-disclosure | Account suspension or termination |
| Previously published AI content, no update | Retroactive flag, update requested |
This escalation matters for publishers who built their KDP business before 2023 and may have a catalog that predates the disclosure requirement. Proactively auditing older titles and updating disclosure status reduces the risk of retroactive enforcement actions.
What this means for governance teams
If your organization publishes content through KDP, including employee handbooks, training guides, marketing materials, or any commercial content, the enforcement escalation has direct implications for your AI governance documentation requirements.
The risk profile has three components:
Documentation risk. Amazon's enforcement depends on what you disclosed, when. If your publishing process uses AI tools but your KDP records show no AI disclosure, that gap is the liability. The fix is procedural: build AI disclosure into the publishing workflow so every title that uses AI assistance gets correctly flagged at submission.
Copyright risk. AI-generated content occupies uncertain ground under US copyright law. The Copyright Office has declined to register works produced entirely by AI without human authorship. For publishers, this means AI-generated titles may not carry the same copyright protection as human-authored work. For a full analysis, see AI output copyright risk for commercial use and generative AI copyright ownership in 2026.
Training data risk. The books you publish on KDP may be used to train future AI models. For publishers working with proprietary content, understanding the data rights implications of KDP's terms of service matters. The broader legal picture on AI training data copyright fair use is still developing.
Compliance checklist before your next KDP publication
Before publishing any title that involved AI tools in its creation:
- Determine which disclosure category applies: AI-generated text, AI-generated images, AI-generated translation, or AI-assisted (where AI helped with editing, research, or ideation but did not generate the substantive content)
- Check the correct box in the KDP publishing workflow (this is the one mandatory disclosure step)
- Document which AI tools were used and in what capacity in your internal records
- Review any previously published titles for disclosure accuracy, particularly those published before 2023
- If publishing at high volume, audit your submission velocity and category spread (these patterns trigger automated review)
- Review cover art separately from text, since AI-generated images require their own disclosure
- If you use AI for translation, disclose this even if a human reviewed the translation
- Retain documentation of your disclosure decisions in case of an account review
The AI governance checklist for small teams covers broader AI documentation requirements across publishing and other business functions.
What is not changing
Amazon does not flag AI-disclosed titles as lower quality to buyers. The disclosure appears nowhere on the product page. Authors who correctly disclose AI involvement are complying with policy and face no commercial penalty from Amazon's platform mechanics.
The enforcement risk is specifically and only non-disclosure. Correctly disclosed AI-generated content is permitted. The compliance question is documentation, not the use of AI itself.
