Search for almost any nonfiction topic on Amazon right now and something odd happens. You scroll past a dozen books with nearly identical covers, suspiciously round page counts, and author names you cannot find anywhere else on the internet. Click on one and the description is grammatically fine but somehow says nothing. The reviews are either glowing five stars posted within 48 hours of publication or furious one stars from readers who feel cheated. You close the tab and try a different search. Same thing.
This is what shopping for books looks like in 2026. Amazon's catalog has been flooded with AI generated titles at a pace no human moderation team could realistically manage. Estimates from the Authors Guild put the number of AI generated books uploaded to Amazon's Kindle Direct Publishing platform at over 700 per day as of late 2025. That number has almost certainly grown. The effect on real authors trying to get their work noticed is not subtle anymore. It is the central problem in self publishing right now, and it is starting to bleed into traditionally published books too.
How Bad Is It, Exactly
Bad enough that writers who have been publishing on Amazon for years are reporting the worst discovery numbers they have ever seen. A romance author who asked not to be named said her monthly page reads through Kindle Unlimited dropped by roughly 40 percent between mid 2024 and early 2025, with no change in her publishing schedule or marketing. She blamed the algorithm. Her books were getting buried under a wave of AI generated romance titles that were being uploaded sometimes 50 at a time from single accounts.
The mechanics of why this happens are pretty straightforward. Amazon's recommendation algorithm rewards books that get purchased or borrowed quickly after upload. Content farms figured this out. They upload a book, flood it with fake reviews to trigger early sales signals, and the algorithm starts recommending it. That recommendation slot is now occupied. The real author's book, which took eight months to write and went through three rounds of editing, is competing for the same slot and losing.
It gets worse in certain categories. True crime, self help, financial advice, children's books, and anything adjacent to a recent news event are particularly saturated. Search for a book about a trending true crime case and you will find AI generated titles from accounts that did not exist two weeks ago sitting above books from established true crime writers with years of research behind them.
What the Platforms Say vs What They Do
Amazon updated its KDP content guidelines in 2023 to require authors to declare if their content was AI generated. The policy is real. The enforcement is a checkbox. Nobody at Amazon reviews a manuscript before it goes live. The declaration system runs entirely on honesty, and the people flooding the platform with AI content are not particularly honest.
Apple Books and Kobo have faced similar problems, though at lower volume. Barnes and Noble Press introduced stricter account verification for new publishers in 2024, which helped somewhat, but the majority of the problem sits on Amazon simply because Amazon is where the money is. Kindle Unlimited alone pays out over half a billion dollars a year in author royalties, which makes it an obvious target for anyone trying to game page reads at scale.
In March 2025, Amazon announced it would begin using AI detection tools to flag potentially AI generated content before publication. Authors who tested the system found it inconsistent. Some clearly AI generated books sailed through. Some human written books got flagged. Amazon has not published specifics on how the system works or what its accuracy rate is. The Authors Guild called the announcement a step in the right direction and not nearly enough.
Goodreads, which Amazon owns, has a separate but related problem. Fake books that exist on Amazon automatically get Goodreads pages. Readers who do not realize a book is AI generated leave reviews there too. Some of those reviews end up on the pages of legitimate books by real authors with similar titles. The cleanup process is manual and slow.
The Real Cost Nobody Talks About
The financial hit to individual authors is real but hard to pin down precisely. The less visible cost is what happens to readers who get burned a few times and start trusting the platform less. A reader who pays five dollars for an AI generated book that turns out to be a reformatted Wikipedia article does not just get a refund and move on. They get more skeptical. They start relying more on personal recommendations and less on platform search. That is bad for every author on the platform, including the ones writing genuinely good books.
Librarians have started flagging this problem from a different angle. Several public library systems that offer digital lending through services like Hoopla or OverDrive have reported that patrons are returning books unfinished at higher rates than in previous years. Not all of that is attributable to AI content, but librarians in communities that have discussed the issue say readers are increasingly frustrated with books that feel hollow. They borrowed it because the cover looked professional and the description sounded interesting. Then they opened it and found something that read like it was assembled rather than written.
What Some Authors Are Doing About It
Some writers have started fighting back in practical ways. A group of self published romance authors created a shared spreadsheet in 2024 to track known content farm accounts on Amazon. They report the accounts collectively, which gets faster responses than individual reports. Several accounts with hundreds of AI generated titles have been removed as a result, though new ones appear regularly.
Other authors are leaning harder into personal branding as a way to survive the flood. If readers follow you specifically rather than searching by genre, the algorithm matters less. Newsletter lists, author websites, and direct reader communities on platforms like Substack or Patreon have become more important to midlist authors than they were three years ago. It is more work on top of the writing itself, which is already a full time job, but authors who have built direct reader relationships say those readers are not going anywhere.
There is also a growing conversation among authors about unionizing their response in some form, pushing for platform level changes through collective pressure rather than individual complaints. The Authors Guild has been the loudest institutional voice, but independent communities of self published writers are organizing in ways that did not exist before this problem became impossible to ignore.
🛡️ How Readers Can Actually Help Right Now
- Report AI generated books directly on the listing page. Amazon has a report button on every product page. Use it. Enough reports from real readers get human review faster than author complaints alone.
- Leave honest reviews that name the problem. A review that says the book appears to be AI generated, with specific examples like repeated phrases or factual errors, helps other readers and creates a paper trail that platforms cannot easily ignore.
- Follow authors directly, not just genres. Subscribe to newsletters from authors you like. Follow them on social media. When you discover a book you love, the best thing you can do is tell someone, because the algorithm probably will not.
- Check before you buy. Search the author name outside of Amazon. Do they have a website? Other books with real publication histories? Interviews or reviews in places other than their own Amazon page? Two minutes of checking can save you money and frustration.
- Buy from authors when you can. Some authors now sell direct through their own websites using tools like Payhip or Gumroad. The author keeps more of the money and the sale happens entirely outside the platform ecosystem that is causing the problem.
The honest truth is that the platforms are not going to fix this quickly because fixing it costs money and slows down volume. Amazon makes money on every book sold regardless of whether a human wrote it. Until that calculus changes, either through regulation, meaningful financial pressure, or some form of collective action that actually hurts platform revenue, the incentive to clean things up stays weak.
That leaves readers and authors in a strange position together. The readers want good books. The authors want to be found. The platform that connects them is currently more interested in the number of connections than the quality of what gets connected. That is not a new problem in internet businesses, but it is hitting publishing harder right now than almost any other creative industry, and the people who care most about books are the ones paying for it.