In 2023, Amazon capped Kindle Direct Publishing (KDP) authors at three new self-published titles a day, an attempt to slow a rush of low-effort, machine-written books clogging the store. Three years on, according to The Decoder, that flood hasn't slowed — it has grown large enough to visibly dent sales for human authors on the platform.

The pattern described in the report is straightforward: AI tools make it trivial to generate a full-length manuscript, cover art, and back-cover copy in a matter of hours, then upload it to KDP for near-zero marginal cost. Multiply that by however many people are doing it simultaneously, and entire genre categories — romance, self-help, children's picture books — get crowded with titles that didn't exist a year ago.

Writers and publishers quoted in the piece describe a market where discoverability, not just competition, is the real casualty: it's not that AI books are necessarily better, it's that there are so many of them that human-written titles get buried in category rankings and search results.

Why volume beats quality on Kindle

Amazon's storefront ranks books largely by engagement signals — sales velocity, page reads through Kindle Unlimited, review counts, and keyword match. None of those signals require a book to be good; they require a book to move. AI-generated titles are cheap enough to produce that authors (or publishing mills) can:

A human author writing one book a year simply cannot compete with that cadence on a ranking system built around volume and velocity, even before questions of writing quality enter the picture.

Amazon's policy hasn't kept pace

Amazon does require authors to disclose AI-generated content in KDP, and it does cap daily uploads. But the report notes those guardrails haven't been enough to stop the practice at scale — disclosure is self-reported and lightly enforced, and a three-book-a-day cap still allows over a thousand titles a year per account, let alone across the many accounts a single publishing operation can run.

That gap between stated policy and actual enforcement is the part worth watching. It mirrors a pattern seen across other AI-flooded platforms — app stores, stock photo marketplaces, YouTube — where a rule exists on paper but the review pipeline can't scale to match submission volume.

What this means if you're building on AI-generated content

For teams building products around AI-generated text, images, or media, the Kindle situation is a preview of a problem that shows up anywhere a marketplace ranks by volume:

The underlying tension — cheap generation versus scarce human attention — isn't specific to publishing. Any product that lowers the cost of producing content to near zero will eventually need a distribution mechanism that doesn't simply reward whoever produces the most.

AiiN's takeaway

This isn't an argument against generative tools for writing — plenty of legitimate authors use AI for drafting, editing, or research. The problem is structural: Amazon built a discovery system optimized for an era when publishing a book was expensive, and generative AI broke that assumption overnight. In our estimation, platforms that rank content by volume or velocity will face versions of this same flooding problem well before they get around to fixing detection or disclosure — Kindle is likely just the most visible early case because publishing has an unusually direct link between rankings and someone's paycheck.

For AI builders, the lesson isn't to stop generating content at scale — it's that whatever marketplace or distribution layer sits downstream of your generation pipeline needs abuse controls built for AI-era volume, not pre-AI assumptions about how fast humans could produce.