AI Fiction Hits Amazon's Top Sales Ranks, Undisclosed

Researchers matched AI-detection scans of 14,419 self-published Amazon novels to sales data through June 2026, per an arXiv paper. Books with more than 25% AI-detected text are landing in the scarce top sales ranks once held by human-written titles. None of them disclose AI content to buyers.

Over the period, quarterly sales volume across these titles grew 19.2 times, but quarterly revenue grew only 8.9 times. That gap means revenue per selling book fell across most genres. The drop is sharpest in genres with heavy AI diffusion and where Kindle Unlimited availability is high.

The authors don't chalk this up to these books being better written. They call it dilution through scale, more books chasing a slower-growing pool of money. They tie the pattern directly to the market-effect prong of fair use, the test for whether a use damages the market for the original. It's the first dataset I've seen built specifically to answer that question with real sales numbers instead of anecdotes.

Amazon doesn't need to ban AI fiction to fix the ranking distortion. A disclosure requirement alone would likely reshuffle the top ranks within a quarter, because what's winning here is volume, not quality. I'd watch whether other AI detection efforts start pairing text analysis with actual revenue data instead of stopping at detection rates.