← Back to blog

KDP's AI Disclosure Rules: What They Cover, What They Don't, and What's Next

Amazon asks if you used AI but never tells your readers the answer — and that gap is reshaping indie publishing.

By Vlada Matusova

Amazon KDP now requires authors to disclose whether AI-generated or AI-assisted content was used in their books — but here's the uncomfortable truth: that disclosure stays hidden from every single customer who clicks "Buy Now." Despite pressure from organizations like the Authors Guild, which directly tried to persuade Amazon to make AI disclosure public, the retailer still does not surface that information to readers. That means KDP's policy functions less as consumer transparency and more as an internal liability shield — one that protects the platform while leaving authors to navigate the reputational fallout on their own.

Let's be specific about what the rules actually cover. When you publish through KDP, you must indicate whether your text, images, or translations were produced with the help of generative AI tools. Amazon distinguishes between AI-generated content (where AI produced the material) and AI-assisted content (where AI served as an editorial or formatting aid that a human substantially modified). If you used Midjourney for your cover or ChatGPT to draft chapters, you're expected to flag it. Failure to disclose can result in takedowns, and we've already seen traditional publishers act on this: Hachette pulled a novel over undisclosed AI use, sending a signal that the industry is not treating this casually. But KDP's enforcement mechanisms remain opaque. There's no public audit trail, no verification system, and no penalty structure that's been transparently communicated to authors. The rules exist, but the teeth behind them are blunt.

What the rules conspicuously don't cover is the thing that matters most to indie authors building a loyal readership: buyer-facing transparency. If you're an author who writes every word yourself, KDP gives you no way to distinguish your work from an AI-generated title in the marketplace. Your disclosure and the AI-heavy author's disclosure land in the same invisible backend field. This is precisely the frustration that led the Authors Guild to launch a Human Authored certification mark — a pending registered certification trademark with the US Patent and Trademark Office, modeled after consumer trust signals like Fair Trade or Energy Star labels. The Guild developed it because, as their leadership explained, "some authors who are not [using AI] want a way to stand behind their work." It's a licensed trademark program, not a logo anyone can slap on a cover, and it represents a market-driven response to a gap Amazon has refused to close.

The legal landscape adds another layer of uncertainty. Federal courts have issued rulings on AI and copyright that, so far, have given authors neither the clear protections nor the definitive precedents they need. One federal judge ruled that AI model training constitutes fair use — with a significant caveat that's still being parsed by legal scholars — while other cases remain in progress. The question of whether you can copyright a work created with AI assistance is still being litigated and debated. For indie authors, this means the content you disclose as AI-assisted on KDP may not enjoy the same copyright protections as fully human-authored work, which has direct implications for your IP strategy, licensing deals, and long-term catalog value.

Meanwhile, the self-publishing infrastructure itself is buckling under these pressures. Writers and publishing consultants have reported that the KDP and IngramSpark experience has deteriorated significantly, with new friction that appears driven in part by platform-level pushback against AI-generated books and tightened identity verification. One experienced publishing coach described the current landscape as having "changed drastically, for the worse," citing weeks of struggles with processes that used to be straightforward. For authors with one to three years of experience, this means the rules are shifting underneath you — not just what you must disclose, but how smoothly your legitimate, human-written work moves through the pipeline.

Here's my position: KDP's disclosure rules are a floor, not a ceiling, and treating them as sufficient is a strategic mistake. The authors who will build durable reader trust over the next five years are the ones who go beyond backend checkboxes and proactively signal their creative process. The one concrete step you should take this week: visit the Authors Guild's Human Authored certification page, review the application requirements, and evaluate whether the certification mark is right for your catalog. It costs time and scrutiny, but it gives you something Amazon currently won't — a visible, verifiable promise to your readers that a real person wrote the book they're holding.