PARTNERSHIPS | COMMUNITY | PODCAST | FRIENDS
AI Companies Are Accused Of Hoarding Rare Books For Training
De-spined and deep-fried // Image by Kait Cunniff with DALL-E
Independent booksellers across the country are reckoning with the fact that increased sales are coming from an unlikely source — AI companies tearing apart books to train LLMs.
The Big Picture: An LLM is only as good as its dataset. As the top firms race to gobble up as much information as they can, the ability to track down and ingest even the most niche topics could theoretically give companies an edge in answering any user query. But that quest for knowledge is already ridding the world of a century’s worth of physical books. Is the cost worth it?
Between The Pages: Indie book shops all reported a similar phenomenon to WSJ — bulk book orders are literally disappearing into the cloud.
The orders are typically placed through third-party marketplaces like Alibris, Biblio, and Amazon by unusual buyers like Red Sparrow Project, Blue Finch Project, and Green Parrot Project.
They typically order hundreds of obscure titles at a time, like Home Long-Term Oxygen Treatment in Italy and Marriage Litigation in Medieval England — especially ones with an ISBN (a unique identifier given to published books since 1960).
Suspicious sellers have tracked these books to facilities “where books are sliced up, scanned, and destroyed.” Taking books apart makes them easier to scan than bending the spine.
They’re then sent to recycling facilities, where the books become your next iced-coffee straw or roll of toilet paper.
Last Chapter: Booksellers are very conflicted about AI companies’ practices. On one hand, revenue at some of these shops has doubled over the past few years, and many of the books sold are considered “dead stock” — titles that were heading for the recycling bin anyway.
But booksellers are concerned that rare titles are getting lost in the shuffle, putting access to them solely in the hands of AI companies. Anthropic says it doesn’t destroy antique books, while SpaceXAI says it created a library to “preserve any rare books.” Still, some sellers are holding back titles from bulk buys or asking other sellers to fulfill orders if the only copies they have are signed by the author.
The Future: Don’t be surprised if indie booksellers organize to create a consortium that logs and protects endangered books from finding their way into an AI firm’s shredder.
Together with SuperhumanAI
Go From AI-Overwhelmed To AI-Savvy Professional
AI keeps coming up at work, but staying on top of it is easier said than done.
That’s exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.
Here’s what you get:
Daily AI news that matters for your career — Filtered from 1,000s of sources, so you know what affects your industry.
Step-by-step tutorials you can use immediately — Real prompts and workflows that solve actual business problems.
New AI tools tested and reviewed — Everything is vetted to surface the tools that drive real results.
All in just 3 minutes a day.
Today’s email was written by David Vendrell.
Edited by Nick Comney. Polled and Copy-edited by Kait Cunniff.
Published by Darline Salazar.