#destructivescanning — Public Fediverse posts

We Tracked a Shipment of Rare Books. It Ended at an Amazon AI Training Facility

https://www.404media.co/we-tracked-a-shipment-of-rare-books-it-ended-at-an-amazon-ai-training-facility/

https://archive.is/rs4aE

#ai #books #datacenters #rarebooks #tech #technology #aitraining #generativeAI #amazon #culture #destructiveScanning

@techwire #destructivescanning

"Destructive scanning was Anthropic’s solution to a problem: in order to “train” Claude, Anthropic had to procure a large, high-quality language dataset, preferably one created before 2022 and the corrupting influence of generative AI on contemporary text."

#DestructiveScanning #books #BookLife
#AI #Claude #Anthropic

https://www.theguardian.com/commentisfree/2026/aug/05/anthropic-ai-destroying-books

Scan everything. Destroy nothing.

So we’ve officially reached the stage of history that Orwell and Bradbury wrote as warnings.

We’re destroying books.

Not burning them, to be precise this is the twenty-first century and we have more sophisticated methods now. We’re scanning them, training AI models on their content and then, in some cases, putting the physical copies in the bin. Rare, antique and out-of-print books to be precise. Apparently coined as destructive scanning, which is the kind of bureaucratic language that makes something genuinely alarming sound almost reasonable.

I assume Bradbury would have thoughts.

I should say upfront that I understand the scanning aspect. Digitising books isn’t new; in fact, digitisation is one of the better things the internet has given us with balance. It means someone on the other side of the world can read a manuscript they would otherwise never have access to; fragile texts are protected from constant handling; it preserves knowledge against floods, fires and simple bad luck. I wholeheartedly believe if there’s a rare manuscript sitting in an archive somewhere that can be read by thousands of people without anyone having to turn another page, that’s extraordinary.

What I can’t get my head around is the destruction. I’m also far less comfortable with the idea of training commercial AI systems on other people’s work without meaningful permission or compensation but that’s a separate ethical question. I just can’t shake the feeling that destroying a rare book after scanning it says something uncomfortable about how we value physical objects in a digital age.

Perhaps the economist would tell me I’m being too sentimental. Storage costs money, don’t you know and rare books are expensive to maintain. Once you’ve captured the text, the information survives, so what’s the problem?

But is a book only its text?

I don’t think it is.

A first edition isn’t interchangeable with a reprint, even when every word is identical. Charles Darwin’s notebooks arent fascinating because we don’t know what he wrote; we do, we have done for over a century. They’re fascinating because they show how he thought. The crossings out, the sketches in the margins, the false starts and the question circles and the ideas abandoned halfway through a sentence. Remove the object and you’ve preserved the information while losing something much harder to define.

Maybe that’s just the book collector in me or maybe it’s because I’ve spent enough time in museums to know that objects carry history in a way a PDF never will. We don’t preserve Roman coins because we’ve forgotten what they look like. We preserve them because the object itself tells a story the image cannot. The same is true of letters, diaries, notebooks and novels. They’re cultural artefacts more than being just decorative; they’re evidence.

The mindset unsettles me just as much as the technology.

Somewhere along the way we became comfortable treating the physical world as temporary, provided we’d made a digital copy first. We photograph paintings rather than looking at them. We assumed the digital version is a perfect substitute for the original when, often, quietly in ways that only become obvious later, it isn’t.

The irony that we are building artificial intelligence by consuming centuries of human knowledge whilst simultaneously destroying the physical record that carried that knowledge in the first place is difficult to ignore. We are training machines to think like humans by eliminating some of the evidence of how humans actually thought (wtf!).

Libraries have spent centuries fighting to preserve books. Wars have been fought around them. Monks copied manuscripts by hand because they believed the ideas were worth the labour. Archivists spend entire careers trying to stop paper from disintegrating into dust. So it feels oddly backward to arrive in the twenty-first century, with all the technology at our disposal and conclude that the book itself no longer matters.

Maybe I’m overreacting; perhaps this is an acceptable trade-off that I’m sure plenty of people far more knowledgeable than me would argue that it is. Maybe this is simply what progress looks like and I need to make peace with it.

or maybe the friction comes not from whether AI should learn from books. Of course it should; books are one of humanity’s greatest achievements and also, as it turns out, a fairly comprehensive record of everything we’ve thought and done, everything we’ve got right and got wrong. The friction is why preserving knowledge suddenly requires destroying the objects that carried it here.

I’ve been keeping up with this story across socials, adding my own thoughts and watching people respond with versions of “well that’s progress”, “it’s inevitable” as though progress as ever been an argument on its own.

The books being destroyed right now are not coming back. This is not a recoverable situation. It’s been reported it’s happening on a huge scale and the question of who decided that was an acceptable trade-off and why it happened quietly, without much resistance is going to matter considerably more than it currently appears to.

This is dystopian. I don’t think that word is an overreaction. I think it’s accurate.

Thank you for reading Extracurricular! Subscribe for more thoughts collected between books, uni and research rabbit holes ★