Comment by _djo_
7 days ago
There are a ton of niche books like the one the parent is speaking of which might not be considered hugely valuable by the general public or less specialised second-hand stores, but are both rare and valuable in their niche.
For instance there were only limited production runs for a set of books on South Africa's participation in the Second World War, and getting hold of them is increasingly difficult. But they have ISBN numbers, so they fall within the group of books being collected and destroyed by AI companies, and as a result might disappear altogether, along with the knowledge inside them.
> There are a ton of niche books like the one the parent is speaking of which might not be considered hugely valuable by the general public or less specialised second-hand stores, but are both rare and valuable in their niche.
I'll ask again: Does anyone have a single example of this?
Every conversation about this topic has been full of this same assertion, but nobody can ever name even a single example. We're supposed to believe it's factual.
You’re supposed to believe it’s factual because you have investigative reporting by multiple newspapers with good track records, on the record statements by bookstore owners, and evidence presented in court cases like the one regarding Anthropic’s Project Panama.
Notably, one of those piece of evidence from that Anthropic case was an internal presentation clearly stating the intention of Project Panama was to “destructively scan all the books in the world.”
Beyond that, some book sellers have named specific titles, such as “an Estonian translation of John le Carré's The Mission Song, a specific edition of Anne Brontë's Agnes Grey, and the October 1983 issue of Warship monthly magazine” and “everything from 18th-century African agricultural implements to biographies of 1950s racing drivers.”[0]
But getting a full list of all those sold globally is difficult. I imagine we’ll start to see some efforts to try to encourage booksellers to contribute to public lists though.
[0] https://www.theguardian.com/technology/2026/aug/15/uk-irelan...
Seems like getting a copy and scanning it and releasing it as part of a digital library would be a good way to preserve the knowledge.
Making the contents of the books available verbatim afterward would be a less bad option, yes.
That's not happening though.
Right, but that's because of copyright law.
3 replies →