Anthropic judge rejects $1.5B AI copyright settlement

Posted by nobody9999 9/9/2025

Anthropic judge rejects $1.5B AI copyright settlement(news.bloomberglaw.com)

293 points | 320 commentspage 2

nobody9999 9/9/2025|

xvector 9/10/2025||

The tech companies need to make a shared subsidiary that simply buys all the books just once and then shares that data amongst subsidiary owners.

internet_points 9/10/2025|

Buying a book at the regular price doesn't give you rights to sell copies of that book.

(I'm not saying selling llm access means selling copies of the book -- but then I'm also not not saying it.)

HardCodedBias 9/9/2025||

Outputs not inputs needs to become law.

freejazz 9/10/2025||

This headline is incorrect.

throwmeaway222 9/9/2025||

[flagged]

jawns 9/9/2025||

In class action lawsuits (Rule 23, Federal Rules of Civil Procedure in the U.S.) any settlement must be reviewed and approved by the judge, who has a duty to ensure it is "fair, reasonable, and adequate" for all members of the class. Judges often reject settlements if they appear unfair, too favorable to one side (often the defendant), or if attorney fees seem excessive.

SilverElfin 9/9/2025|||

They may not have a choice. Fighting these legal battles is very expensive and exhausting. And books are a low margin business. Anthropic has access to funding. Authors? Not so much. Losing your book to AI for a one time $3000 settlement feels like a bad deal to me.

pier25 9/9/2025|||

> Authors? Not so much. Losing your book to AI for a one time $3000 settlement feels like a bad deal to me.

AFAIK it's even worse as this settlement is only about downloading pirated copies of the books. IIRC the training itself was deemed fair use.

Dylan16807 9/9/2025||

Well things are worse in the sense that a different ruling feels really bad for them.

But it means that this case is getting them thousands of dollars instead of one or two purchases, which is a pretty good outcome.

visarga 9/9/2025|||

> Losing your book to AI for a one time $3000 settlement feels like a bad deal to me.

I'm wondering how lost would the book be? What would be the difference in sales.

program_whiz 9/9/2025||

It isn't about sales in the short term. Same with code. Your project might even be OSS (so not expecting profit). Its about a system that exploits that for one party's profit, while putting the other out of business. They reduced the writer's ability to ever be employed again, or to make royalties at the same level as before (not necessarily reduced the sales of existing titles, though that may occur as well).

Same with code, AI hoovering all the code doesn't mean people won't use libCurl, but it does mean jobs are disappearing and people may not be around to write the next libCurl.

ares623 9/9/2025||

How does the saying go again? “Everyone’s a libertarian until they get punched in the face.”

NelsonMinar 9/9/2025|||

Incorrect information from a throwaway account. This is simply not true.

dboreham 9/9/2025||

Very wrong.

ceejayoz 9/9/2025||

Gotta love the confidence, though!

atleastoptimal 9/9/2025|

Copying files of scanned books isn’t worth a 1T dollar fine

adrr 9/9/2025|

Maybe Anthropic should have paid attention to the law that has $150k statutory damages per violation if the infringement is willful. So much cheaper just to buy the books and scan it instead of violating a law that has a statutory damage clause.

ripped_britches 9/9/2025|||

I really want to see an alternative universe where we have mechanical turk folks scanning a huge literal book library into a data warehouse.

kg 9/9/2025|||

Google and the Internet Archive both did/do this with elaborate setups. https://archive.org/details/eliza-digitizing-book_202107

wrsh07 9/9/2025|||

I think anthropic has this operation now

atleastoptimal 9/10/2025||||

If a person pirates a book should they have to pay 150k?

ch_fr 9/10/2025||

So true! I too think the average person is basically indistinguishable from anthropic.

atleastoptimal 9/12/2025||

And I remember the part in copyright law where it says "if the violation is done by a big corporationy corportation, then the punishment must be more severe even on a per-violation basis"

ch_fr 9/12/2025||

I mean, it's not crazy talk, you frame it as crazy but this difference in treatment is fairly normal. If punitive damages were limited to what an individual person could manage, they would hardly ever be "punitive" for corporate actors.

You're the second person I've talked to in this thread who thinks "The law is not applied with perfect consistency in practice!? Dear god why is it not like computer programs?".

It's just my personal take, but I don't think extreme rigidity and consistency in a "code is law" fashion would ever be desirable. look over to the crypto world to see how DAOs are faring.

atleastoptimal 9/12/2025||

150k per violation of copyright law by means of pirating a book is excessive even for the largest company in the world. I agree that punishment should be somewhat proportional so that "too big to fail" firms don't see it as just a fine to be accounted for as any expense, but among things that ought to be worth a a fine of that magnitude, downloading a book is not among them.

ch_fr 9/13/2025||

And I agree that 150k is dialing the slider a bit too high compared to an individual's offense, happy to meet you at the middle there.

eschaton 9/9/2025|||

In general buying books and scanning them for this type of use would *also* be copyright infringement.

adrr 9/9/2025||

No. Thats fair use. Format shifting is fair use as affirmed by RIAA v. Diamond Multimedia which was about ripping CDs to MP3s.

eschaton 9/9/2025||

…for personal use, just as timeshifting was with MPAA v. Sony. Neither were about commercial use/exploitation.

adrr 9/9/2025||

Meta case just affirmed that training an LLM is fair use under transformative use. Alsom, Google's indexing(transformative use) of scanned books is settled law with Authors Guild v. Google.

eschaton 9/10/2025||

And Roe v. Wade was settled law too, until it wasn’t.