Just like
the Nazis did.
Once done with them, why don't they just give them away ffs.
AI companies are buying loads of physical books, hoovering up the texts for model training, and then physically destroying the originals. A group of 18 advocacy organizations on Friday asked the US Federal Trade Commission to investigate the book butchering and knowledge hoarding. Fresh details of the practice emerged earlier …
> Once done with them, why don't they just give them away ffs.
Because the cheap/fast way that they are scanning them involves chucking them into a hopper, wherein a machine gruesomely slices off the spine with a saw, and then the pages are fed to an industrial auto-feed scanner, similar to the type seen on the top of office photocopiers.
What's left is a pile of unsorted pages and sliced-off spines which are of little use to anyone, so they either pulp them or burn them.
The problem is, these are -rare books- which may be considered irreplaceable historical artefacts. The scanning process is crude and low-quality, there will be missing/unreadable pages in the output, and there will be details that are lost on the OCR process.
It's not even clear that they are even keeping the raw data after running their slop-training. At the very very least, they should be required to make any out-of-copyright books available as PDF to the public.
But they would never do that because it would destroy their "commercial advantage". Perhaps they should pool their efforts into a nonprofit that makes it available to all. The last thing we need is multiple competing AI labs destroying multiple copies of the same rare books to try to get an edge over eachother.
I suspect he might be down the bottom of the thread.
There are some people who quasi-religiously, death-cultishly worship the AI companies. They can do no wrong, so long as they bring the Absolute1 forth into the world, it matters not that all history and verifiable fact is consumed.
1 Thanks to Bebu Sa Ware for that recommendation! Yet another alarmingly prescient warning from the 1920s.. An amusing, enlightening and horrifying read.
This post has been deleted by its author
They also benfit by removing the text from the available pool of words.
If your competitors can't scan it, your model may have a unique/larger basis of text.
The philosophy applied here is driven by the purest desire for information hoarding.
Humans wrote the literature, these jokers ingest it for their exclusive private ownership... And then rent the outputs back to humans... Fuck them.
You do know Hamas was put into power to stop the two state solution by your current leader using funds given to them from the UAE by his command. Google it while you can still see it,
This is the exact point I'm making about history.
Why don't you explain why you keep voting in genocidal maniacs that commit war crimes much worse than any war crimes we have seen throughout history?
Hamas vote was decades ago with no elections since. Yours was not. Check yourself out.
I really, really hate trolls on the internet you Hasbara coward.
I'm no fan of Kier Starmer, but I wouldn't describe him as a genocidal maniac, maybe you're confusing me with someone else? Hasbara is certainly not an applicable label.
In the last Palestinian election Hamas and Fatah had similar shares of the vote, but Hamas ended up with many more seats. Even Fatah gave up on them eventually because of their violent behaviour. It will be interesting to see what the November election results are, will a majority of Palestinians vote for peace, or violence?
Don't claim to be someone you are not. Kier Starmer isn't even in power any more so you need to update your Hasbara bots.
Is this what AI is being used for now?
I tell you what. If I had the chance the first books I would burn and I would burn every single copy are the Talmud, the Bible (old and new testaments) and the Quran. Some religious people can be so hateful.
Well, I mean as a named place, and an identifiable geography, it only existed a few decades, and has been gone since before Yugoslavia. How often do you hear that name?
Gaza and the West Bank are identifiable places. I'm not trying to deny they exist. Realistically though, they're not a place called Palestine, certainly they're not the place the British named Palestine. Maps change. Always have and always will.
I think a relevant fact most posters are overlooking in the thrust of the thread is that the pulp fiction quantity books are being destroyed but the rare books are not. At least according to wide spread reporting elsewhere.
No, these were purchases by the AI companies where the books, rare and otherwise, were all thrown into the spine slicing and scanning process. They just want every book, heedless of the damage. See https://www.theguardian.com/technology/2026/aug/02/australian-book-sellers-alarm-destruction-rare-titles-ai-supply-chain for example, similar stories have run on the BBC.
Scanning and then destroying the original texts to save storing them? Hypothetically would that include the original copy of your Declaration of Independence, the earliest Bibles etc? After all who cares if you can see Thomas Jefferson’s actually writen signature if it helps Anthropic’s* share price?
* other AI companies are available.
> The book no longer exists after it's been scanned
The crude and lazy way these companies are doing it, true.
Of course, it *is* possible to scan a book without damaging it - in fact, to preserve it from wear and tear whilst making its contents accessible to more people.
But in order to do that you have to spend, ooh, millions of dollars on the project and the AI companies are completely strapped for cash, they could never afford to do buy/make one of those robotic scanners that takes high-res images. Which images include data about the way the book was created as well as just allowing the characters to be read. Because what could you possibly use that kind of data for? I mean, you'd have to feed it into some kind of machine to compare and contrast with other books and manuscripts and where could we ever find a machine big enough to do that?
Morty: Are we gonna put it back when we're done with it?
Rick: Why? The only thing of value on it is the secret treasure map. The rest is just instructions for running a country. I'm pretty sure they're online.
Morty: It just feels a little wrong.
Rick; Morty, are you gonna be a fucking America nerd or are you gonna be cool and steal the Constitution with Grandpa?
(Morty then proceeds to accidentally destroy the constitution with an energy beam, and incidentally also demolish the Washington Monument, explode the Liberty Bell, and release the steampunk Trojan mech the French had hidden inside the Statue of Liberty.)
--- is criminal.
I own something like 500+ books including about 300+ technical and history books, some of which, although not valuable, are quite rare. I saved over 100 technical works on fire, explosion, chemical safety and process plant design from a library that was about to be thrown into a skip when a company closed one of its' engineering divisions. I am aware that some of the technical books I have are sufficiently rare that the only copies in the UK (other than mine) are one at a Russell Group university and another in the British Library. Never mind that there are some fiction books which are becoming sufficiently rare to become valuable, I don't have any of those but a friend has a collection of books written in the 18th / 19th century which are stored in a secure, humidity controlled room. I have no idea what they are worth as he doesn't discuss this, but I'm aware that some are exceptionally rare.
Regardless of this, the destruction of books to generate AI slop, is, in my opinion neither justified, nor forgivable.
Generally speaking I agree with you. Book burning is rarely a good thing - I make exceptions for anything by Russel Brand.
Regarding your library, it seems to me the artefacts have two planes of value. One is obviously the books as artefacts. The second is the knowledge they contain and the unique expression of that knowledge.
Rare texts can, and in my opinion should, be digitised for preservation, however there exist nondestructive techniques for this. Your friends humidity controlled room will one day fall to flood or fire or some other destructive force.
That said, as I read elsewhere so the statement isn't my own, how many copies of the da Vinci code by Dan brown does the world need? Books are destroyed by pulping for far less useful purposes than training AI.
First they came for the search engines
And I did not speak out
Because I was not really that arsed
Then they came for the search results with AI.
And I did not speak out
Because I was a bit pissed off but just ignored it.
Then they came for the internet.
And I did not speak out
Because I used a VPN
Then they came for the books.
Slow and steady wins the race. When we are all gone and no one remembers the past it will be written by those in control. They are already doing it with their AI bullshit.
What do we do? Just let our kids get fucked over? Where do we take a stand against this?
Wah! Wah! The LLM companies are paying for a thing and then the thing becomes their property to dispose of as they see fit. Wah!
If the people doing the whining paid the sellers $1 more for these books than the LLM companies are willing to, then they could keep their newly-purchased books in a climate controlled vault for all eternity. But they won't because they don't actually care, and certainly wouldn't spend their own money for the book-preservation cause they claim to care about*. Nah! They just wanna whine.
If the sellers really cared about the fate of their precious books they could choose not to sell them at all, or only sell them with the stipulation that the buyer must keep them in a climate controlled vault for all eternity. But selling with burdensome stipulations would reduce the sale price so it turns out that they don't really care that much either, do they?
I have had enough of these twitter crusaders whose deep and abiding concern extends only so far as writing a whiny tweet or facebook post or whatever but not actually putting in any effort to do anything useful.
* You know how I know? Cause if they cared so much they would have gone out and bought the books they care so deeply about and said books wouldn't be available for the LLM companies to buy.
> Cause if they cared so much they would have gone out and bought the books they care so deeply about and said books wouldn't be available for the LLM companies to buy.
There is a problem here. Most people were completely unaware of the fact that books were being bought wholesale only to be destroyed by slop-mongers. They did it in secret, through contractors and shell-companies, and it was only when a journalist became suspicious and placed an AirTag inside one of the books, that he discovered what they were up to.
If the purchase had been announced in advance, then public libraries and members of the public would have had a chance to save the books, and the AI companies could have been forced to follow better conservation practices for any books that they did buy.
In the same way: You are allowed to buy a listed building, but you are not allowed to demolish it (unless you are the Mad Emperor of the World™). But for the same reason, companies like Wetherspoons frequently set up shell companies to buy historic buildings before they can be listed, only to eviscerate the inside and sell it to Wetherspoons before the locals are aware of what they are up to. Otherwise someone would protest and apply for the building to be listed.
A reference to the Panama Papers ? Which weren't destroyed before their discovery but a great many would have wished they had been.
Realistically these fools are about as shallow as a teaspoon and as bright as a blown 5W incandescent so I imagine the codename was the result of some Friday afternoon C suite spaff session.
sunk before the Hiroshima and Nagasaki bombs is treasured; it is unpolluted by later radiation tests and critical for certain instruments.
And for the same reason, books from before AI slop started to pollute public texts are treasured: they represent the thoughts and word choices of actual humans.
As others have pointed out, destroying books is barbarous as a concept. We'll never know, for example, what was in the library at Alexandria before it burned other than by surviving references to it. The problem, though, is less than books are being destroyed - cutting the spines and covers off so a page reader can scan them is in practical terms the only way to do it efficiently - but that the information they contain is hidden away for ever, rather than being made available. Remember when the Internet was being touted as a place where all humanity's knowledge was quickly and easily available?
One of the joys of a second hand bookshop is finding out-of-print works by authors you may never have heard of. Books which will never appear in a catalogue; books which will never be reprinted; and in particular ephemera - bound copies of magazines, for example. Now we can say goodbye to that, thanks to big tech's selfish acquisition and hording.
This post has been deleted by its author
GitHub Arctic Code Vault -
GitHub captured a snapshot of every active public repository on 02/02/2020 and preserved that data in the Arctic Code Vault.
https://archiveprogram.github.com/arctic-vault/
2nd February 2020 - means the code in that snapshot is not tainted by AI slop
In practicality, now they've been scanned, we should seek global recognition of the idea that as soon as out of copyright, the digital copy for training should be published.
Turn the loss into a win by making a modern day digital library of Alexandria, free for all to use and download.
the people here - all one of the corporate bootlicking bastards would have probably used the same arguments for slavery or use the line I used to hear in the 80s "black people in South Africa are treated SOOOOOOO much better than other countries in Africa"
corporate cap doffers who'd happily lick the shit off their CEOs boot soles while they receive their redundancy notices
We need to actively shun anyone that has anything to do with AI.
If you use it, you are complicit in what these companies are doing. Book burning puts you on a par with the Nazis. You have the ashes of our culture on your hands.
Do not use AI. Do not pay for AI. Do not work on or at a datacentre, Have nothing to do with these people. Do not speak or consort with those who do.
Time to draw a line under these people and what they are doing.
Allow me to inject a little reality here please.
There are now two kinds of business to work in. Those that use AI, and those that don't. Those that don't cannot survive because modern models are good enough to realise significant productivity gains. They just are.
It's no good hiding in the public sector either. The next government, whoever it is after labour, will force the uptake of AI because on practical terms more than half the back office staff in the public sector can already be automated due to lower productivity.
AI is never going away, just as the 80s are never coming back. You just have to make the best of it, and in time you realise the 90s weren't so bad anyway.
AI is only getting better.
Per the article, if you're not already using AI in cyber security, your attacker is. DefCon was extremely heavy AI where just three years ago AI was just a village. Times have changed and you do have to deal with that reality, even if it is unpleasant.
AI produces productivity gains while the bulk of its cost is paid for by the circular accounting of VCs, Nvidia, Mircrosoft et al boiling the oceans and the people it's replacing are the cream of the crop of the outsourcing-far-beyond-the-cheapest-bidder turf wars.
AI is good at replacing or replicating idiots. Every single time, still no exceptions although I have surely tried, that I've read some sort of "well actually" article or comment defending AI, it turns out the person writing it is just not very good at the thing they've discovered the stochastic parrot is better than them at.
Economies which run on idiots will be overtaken by those which aren't.
Or as we used to say, "go away or I will replace you with a very small shell script language model"
When I called a company and a nice female voice answered the phone, it seemed reasonable - until "she" repeated my phone number back as "one billion two hundred thirty-four million, four hundred fifty-six thousand, seven hundred eighty nine". Ah, they put an "AI" in charge of answering the phone. Not doing business with them.
The first book they scanned !
Hardcover. Condition: Fine. Signed Limited Edition. Bradbury, Ray. Fahrenheit 451; illustrated by Joe Mugnaini. New York: Ballantine Books, Inc., (1953). FIRST EDITION, LIMITED EDITION; "This edition of Fahrenheit 451 has been limited to 200 copies. It is specially bound in Johns-Manville Quinterra an asbestos material with exceptional resistance to pyrolysis. Each copy numbered and signed by the author. No. 159. Ray Bradbury [signature]." 8vo - 8" x 5-3/8". Original white Johns-Manville Quinterra over boards with titles printed in red to front cover and spine, with one tiny dark mark to front cover and minor rubbing to a few of the red letters in the word "FAHRENHEIT" on the spine for what is a spectacular copy of the book, with none of the darkening or spotting to the white asbestos that copies of this edition are prone to. 199 pp. Fahrenheit 451 is a dystopian novel about a future American society where books have been outlawed and "firemen" burn any that are found. Two short stories: "The Playground" and "And the Rock Cried Out" follow the novel in this printing of the book. These two short stories were omitted in later printings of Fahrenheit 451; internally the book is bright and clean, the hinges are both strong and the book remains tightly bound. The condition of this book is FINE. THIS IS A STELLAR COPY OF A VERY SCARCE AND DESIRABLE BOOK. Mugnaini, Joe (illustrator). Signed by Author(s).
They don't scan the books and then destroy them, they destroy the books in order to scan them in the cheapest possible manner. Rip off the covers to get a stack of pages that can be fed into a scanner. After that it's already waste material trow it all into the incinerator. As if they couldn't afford to do it it without destroying the book, but how many mega yachts could you accumulate doing things that way.
Plus, don't just hoard the physical items, hoard the knowledge they contain! Make the knowlegde scarce! There must be a few more private islands available in that racket.
Plus, when there are no more physical records, and the ony way to find out what ever happened in history is to stream it from some corpoaration, imagine the power! Did Greedo shoot first? Are the people in E.T. holding guns or radios? There will be no way of knowing, and the answer can change on a whim with a new inter corporational licensing agreement, or maybe the question will be moot because it will be removed entirely.
Please do not click on this link. It is a young american destroying Model T cars because white privelage makes it possible for him to do so and everyone in his circle says cool bro. There should be a version of it that doesn't enrich him with each view but Google enforces copyright so that is not possible. AMURICA FUK YAAAA!!!
This post has been deleted by its author
Reminds me in a terrible way of a long defunct project/dream at Google Books to scan and make freely available every out-of-print book in existence. Unfortunately , the project was shut down by copyright claims.
One of the major challenges was scanning the books efficiently without damaging them, and they developed some kind of special scanner to that end.
As a teenager when I learned about this, I was inspired to become a lawyer so that I could revive the project. Makes me sad to see an evil perversion of it