KDP's AI Disclosure Rule and the Paper Trail You Need

Amazon's KDP rolled out an AI disclosure requirement in late 2023. For two years it ran on the honor system. In 2025 it became an audit trigger.

The rule distinguishes between "AI-generated" content (the AI produced the words, image, or translation in their final form, with minimal human modification) and "AI-assisted" content (a human wrote it, but used AI tools for editing, ideation, or research). Authors only have to disclose AI-generated content. But the burden of proof, when an audit happens, is on the author.

What we're seeing in 2026: KDP audits are more common on titles that have a few specific patterns. Cover art that matches Midjourney's stylistic fingerprint. Translations done in under a week. Prose that has the cadence of a chat completion. None of these is automatically disqualifying. All of them are flags.

IngramSpark's policy is broader and less forgiving. Their Catalog Integrity terms prohibit content created by "automated means" without distinguishing between generated and assisted. An author who used ChatGPT for a chapter outline is technically in scope.

What to keep on file, just in case:

• Drafts. Save your dated manuscript versions. A timeline of revisions in a Word document or a Google Docs version history is the cleanest defense.
• Cover sourcing. If you commissioned art, save the contract and the artist's invoice. If you used a stock site, save the license.
• Translation chain. If your book exists in another language, document who did the translation and when.
• Audiobook narration. Spotify's 2025 update accepted ElevenLabs in 29 languages, but Audible still treats AI narration differently. Keep records of who narrated.

This isn't about avoiding AI. It's about being able to answer the question if KDP asks. The authors getting taken down in 2026 aren't necessarily the ones who used AI most. They're the ones who can't produce a paper trail when asked.