AI Companies Shredding Rare Books for Training Data
AI companies are shredding rare books
AI companies are bulk-buying rare books, scanning them, and shredding the originals to train models, a practice facilitated by ISBNdb. A federal judge ruled this destruction legal under fair use, arguing that eliminating the physical copy ensures only one version exists. This irreversible loss of historical texts, including pre-2022 works free of AI-generated text, is being masked as digital preservation while companies like Anthropic seek to acquire all global books.
You can re-upload a website. You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data.
- squidbeak
I've limited sympathy for the publishers.
It pisses me off to reflect that they can sit on works until copyright expires, keeping them out of print. There's no real need for any of these so-called rare books to be rare while they're under copyright.
And related to this, the books that are in print are mostly only in print in the shittiest way. I often see well-made books from the 17th or 18th centuries which are still in good nick. It's ridiculous that in the 21st century, publication standards have fallen to the point where for most works a disposable format is the only type available - where no amount of money could buy a truly decent hardback copy.
If we have to have copyright laws, I'd like to see two changes to them.
When a publisher has no incentive to keep an edition in print, it should be available to any other publisher to print, without compensation to the original publisher, and with renegotiated royalties for the author.
And if the publisher keeps a book in print - but only in bestseller-grade materials, bogroll paper that furrows in any humidity and perfect binding that molts its pages a couple of dry seasons later - and if it refuses to print a durable hardback copy with signatures, good paper and decent print - something that will still be readable in several generations' time - any other publisher keen to have a crack at it should be able to, again without any compensation for the original publisher, though perhaps in this case, with matching royalties for the […]
- trollbridge
We reprint old books after checking out copyrights (for all books, this means pre-1930, but for some (I'd actually say most) it also means ones published up to 1964 and 1973, depending on how the rightsholders did (or didn't) do the renewals).
We use a special guillotine type cutter to cut off the binding and then store the pages in a sealed plastic bag which goes in the archives; they're stored there indefinitely in case the book needs rescanned for some reason. We also keep the original, uncompressed copies of the books on magnetic disks.
We also go out of our way to try to find rare books published in 1931, 1932, etc. so they are ready to go once the copyright expires.
And no, no AI company has ever come to us and asked to run training on all of our scanned copies.
- est31
> You can reprint a bestseller. You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal. So it's going to accelerate.
Aren't they shredding only the books still under copyright protection? How is an 18th century botanical text still under copyright?
IDK about the shredding, it's not nice, but it's more a problem with copyright law than AI companies.
Scanning books you own should be legal from a copyright point of view, and not require shredding.
Second, one should think about abandoned property provisions for copyright works published more than 50 years ago and in danger of being forgotten: once challenged, either you as the owner have to prove that the work is preserved for future generations (e.g. in various libraries around the world), or you have to authorize further copies, or you give up copyright on the work.
- lousken
That's why archive.org should have never been sued for lending books they had physical copy of. This is the result. Publishers should be more careful what they wish for.
- ACCount37
The publishers sued AI companies for training on shadow library data, hoping to negotiate content deals for big $$$ down the line. Instead, they got analog hole'd.
Turns out that buying an old book for $5 and destructively scanning it for $25 is way cheaper than paying extortion fees to the copyright-mongers.
What I don't buy is it being "rare, precious books". First, they're not after ancient texts - they're after the books that there's still copyright on. Second, when it comes to books, "old" doesn't mean "valuable" - plenty of libraries destroy old books because there's no demand for them, and storage costs you. This is how those scanning companies get books for so cheap.
- sherr
I see mentions of Bradbury's "Fahrenheit 451" in that thread but what this really seems to be mostly like is Vernor Vinge's "shred and scan" factory in his novel "Rainbows End".
- JumpCrisscross
Maybe we can kill two birds with one stone: digitize rare books and reverse the damage from Authors Guild v. Google [1].
Let AI companies do this. But require them to make the digital copies public. Maybe with a multi-year delay, to give the original scanner advantage to doing it.
[1] https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,....
- Cider9986
Can only blame AI companies to a limited extent. This is apparently the legal way to do things because of stupid copyright laws.
Support your local shadow library: https://annas-archive.pk/donate
- Springtime
It seems a key contention of theirs is the possibility that rare books are being destroyed this way, yet the things they cite don't seem to suggest this (based on their paraphrasing), they just throw the following at the end to make it seem like it's occurring to irreplaceable books:
> You can't replace the last three copies of an 18th-century botanical text once someone shreds them for training data. And the judge said it's legal.
Is there evidence of this? Since otherwise they could very well be describing what is only occurring to in-print or non-rare books. (This is a genuine question since their post doesn't shed any light on it.)
- pu_pe
From what I understand, the rare books in question are not some historically relevant medieval manuscripts, but rather some relatively recent books (still under copyright) for which there are few print copies available for purchase.
I am not sure that physically destructing one copy of this type of book to preserve its contents digitally is so bad. Pretty much anything that is still under copyright should be valuable only for its content, not for the physical medium it's printed on.