I think he implies that because one can borrow hypothetically any book for free from a library, one could use them for legal training purposes, so the requirement of having your own copy should be moot

Libraries aren’t just anarchist free for alls they are operating under licensing terms. Google had a big squabble with the university of Illinois Urbana Champaign research library before finally getting permission to scan the books there. Guess what, Google has the full text but books.google.com only shows previews, why is an exercise to the reader literally

Libraries are neither anarchist free for alls nor are they operating under licensing terms with regards to physical books.

They're merely doing what anyone is allowed to with the books that they own, loaning them out, because copyright law doesn't prohibit that, so no license is needed.

Yup. And if Anthropic CEO or whoever wants to drive down to the library and check out 30 books (or whatever the limit is), scan them, and then return them that is their prerogative I guess.

Scanning (copying) is¹ not allowed. Reading is.

What is in a library, you can freely read. Find the most appropriate way. You do not need to have bought the book.

¹(Edit: or /may/ not be allowed, see posts below.)

Scanning is, under the right circumstances, allowed in the US, at least per the Second Circuit appeals court (Connecticut, New York, Vermont): https://en.wikipedia.org/wiki/Authors_Guild%2C_Inc._v._Googl....

They (OpenAI and Anthropic) operate their platform and distributes these copyrighted works outside, where these foreign laws applies

There are no terms and conditions attached to library books beyond copyright law (which says nothing about scanning) and the general premise of being a library (return the book in good condition on time or pay).

Copyright law in the USA may be more liberal about scanning than other jurisdictions (see the parallel comment from gpm), which expressly regulate the amount of copying of material you do not own as an item.

The jurisdictions I'm familiar with all give vague fair use/fair dealing exceptions which would cover some but not all copying (including scanning) with less than clear boundaries.

I'd be interested to know if you knew of one with bright line rules delineating what is and isn't allowed.

> if you knew of one with bright line rules

(I know by practice but not from the letter of the law; to give you details I should do some research and it will take time - if I will manage to I will send you an email, but I doubt I will be able to do it soon. The focus is anyway on western European Countries.)

Scanning in a way that results in a copy of the book being saved is a right reserved to the holder of the copyright

Afaik to scan a book you need to destroy it by cutting the spine so it can feed cleanly into the scanner. Would incur a lot of fines.

Nah, that's just if you want archival-quality scans. "Good enough for OCR" is a much lower bar.

Anthropic hired the books scanning guy from Google for 1M+ usd to do just that (open the binds).

That's what they did. They also destroyed books worth millions in the process.

They didn't think it would be a good idea to re-bind them and distribute it to the library or someone in need.

To be clear, they destructively scanned millions of books which in total were worth millions of dollars.

They did not destroy old, valuable books which individually were worth millions.

https://arstechnica.com/ai/2025/06/anthropic-destroyed-milli...

I really don’t think there’s any demand out there for re-bound used paper books when most books can be had in their real binding for $3 or less. It would cost at least $3 to re-bind, then they’d have to be listed on Amazon marketplace in “Poor condition” where they’d be valued at maybe $0.50 and cost $3 to ship, and they’d take years of warehousing at great expense waiting to be sold.

As for needy people, they already have libraries and an endless stream of books being donated to thrift stores. Nothing of value was lost here.

> Nothing of value was lost here

But then they shouldn't have done that in the first place. It seems like a crime to destroy so many books.

Imagine, 10 more companies come to join the AI race and decide to do the same.

Books are printed in very large quantities, and there isn't infinite warehousing space for them "just in case." Surplus books just get sent straight to recycling all the time to make room for new books. I would be surprised if while this project was running, it represented even 10% of the daily books being destroyed. It's just never been practical to save every book printed forever.

To be fair, a book is fundamentally a wear item. I remember learning how my university library had its own incinerator. After a certain point it makes no sense to have 30 copies of an outdated textbook taking up space in the racks. Same goes for beatup old fiction and what have you. One might think a little urban school or branch library might want some but they too deal with realities of shelf space constraints and would probably prefer that their patrons had materials more current or in better shape.

That being said, I’m sure these companies did not exclusively buy books at the end of their life.

There are book scanners that don't require cutting the spine, though Anthropic doesn't seem to have used that approach.