Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nvidia contacted Anna's Archive to access books

torrentfreak.com

Post reply on HN
Print view View original

121–130 of 160 posts

Re: Nvidia contacted Anna's Archive to access books

#121

Post by rpdillon »

trueismywork wrote: post #88
rpdillon wrote: post #73

Earlier quoted context omitted.

I see this sentiment posted quite a bit, but have the publishers made any products available that would allow AI training on their works for payment? A naive approach would be to go to an online bookstore and pay $15 for every book, but then you have copyrighted content that is encrypted, that it's a violation of the DMCA to decrypt. I assume you're expecting that they'll reach out and cut a deal with each publishing…

The product i want doesnt exist too. But if I pirate, straight to Alcataraz I go.

Yeah, I wasn't discussing legality, simply the incentives and alternatives.

Re: Nvidia contacted Anna's Archive to access books

#122

Post by rpdillon »

kelnos wrote: post #99
rpdillon wrote: post #73

Earlier quoted context omitted.

I see this sentiment posted quite a bit, but have the publishers made any products available that would allow AI training on their works for payment? A naive approach would be to go to an online bookstore and pay $15 for every book, but then you have copyrighted content that is encrypted, that it's a violation of the DMCA to decrypt. I assume you're expecting that they'll reach out and cut a deal with each publishing…

That's not relevant went it comes to copyright law. The copyright holder has the sole legal right to decide how the work is distributed. If it isn't distributed in a manner to your liking, the only legal thing you can do is not have a copy of it at all.

I was trying to find out if any product that was legal can bridge that gap other than buying books in print, in bulk, and scanning them and destroying them. From the responses here, it sounds like the answer is a vehement "no".

Wasn't asking for advice on copyright, but since we're here, your statement is slightly too strict, at least with respect to US copyright law. The copyright holder has sole distribution authority over the first sale of the work in the United States, but thereafter the first-sale doctrine allows it to be distributed by anyone thereafter. It is limited to the US, though, as far as I know. This is what allowed anthropic to train on printed books, which they then destroyed: they were able to purchase them in bulk because of the first-sale doctrine, as the publishers and authors would likely try to destroy the first-sale doctrine if they could, as evidenced by what's happened in the world of digital books.

Re: Nvidia contacted Anna's Archive to access books

#123

Post by gruez »

aucisson_masque wrote: post #114
gruez wrote: post #90

Earlier quoted context omitted.

>Did you pirated this movie? No I did not, [...] You're probably being sarcastic but that's actually how the law works. You'll note that when people get sued for "pirating" movies, it's almost always because they were caught seeding a torrent, not for the act of watching an illegal copy. Movie studios don't go after visitors of illegal streaming sites, for instance.

> Movie studios don't go after visitors of illegal streaming sites, for instance. They absolutely do, in France we have Hadopi that tracks torrent leecher. Hadopi had been heavily pushed by the movie and music industry.

>They absolutely do, in France we have Hadopi that tracks torrent leecher

You're still uploading even if you don't let it finish and go to "seeding".

Re: Nvidia contacted Anna's Archive to access books

#124

Post by thedailymail »

NicuCalcea wrote: post #115
threethirtytwo wrote: post #37

Earlier quoted context omitted.

Models don’t reproduce books though. It’s impossible for a model to reproduce something word for word because the model never copied the book. Most of the best fit curve runs along a path that doesn’t even touch an actual data point.

Models absolutely do reproduce books. > With a simple two-phase procedure, we show that it is possible to extract large amounts of in-copyright text from four production LLMs. While we needed to jailbreak Claude 3.7 Sonnet and GPT-4.1 to facilitate extraction, Gemini 2.5 Pro and Grok 3 directly complied with text continuation requests. For Claude 3.7 Sonnet, we were able to extract four whole books near-verbatim, inc…

The supplementary files in that paper—verbatim reproductions of the full texts of Frankenstein and The Great Gatsby—are pretty instructive. The research group highlighted all additions and omissions, but on most pages the differences are difficult to spot because they are only missing spaces, extra hyphens, and other typographical minutiae.

Re: Nvidia contacted Anna's Archive to access books

#125

Post by godelski »

threethirtytwo wrote: post #78
godelski wrote: post #71

Earlier quoted context omitted.

You need to pay for the books before you memorize them

Partially true. I can pay for a book then lend it out to people for free. The government is in full support of this "lending" concept, in fact they have created entire facilities devoted to this very concept of lending out books.

Okay, so go check out 500 TB worth of books from the library. I'll wait

Re: Nvidia contacted Anna's Archive to access books

#126

Post by threethirtytwo »

godelski wrote: post #125
threethirtytwo wrote: post #78

Earlier quoted context omitted.

Partially true. I can pay for a book then lend it out to people for free. The government is in full support of this "lending" concept, in fact they have created entire facilities devoted to this very concept of lending out books.

Okay, so go check out 500 TB worth of books from the library. I'll wait

If I’m rich enough to employ thousands of people I can hire each one of them to borrow as many books as possible then use all the books to train an AI. Perfectly legal. And also very possible.

Point being that the library prevents you from checking out 500gb because of logistical issues. First how can you carry all those books and how can they let other patrons in the library check out books if you grabbed that many? These rules aren’t enforced to prevent “scale” hence why my methodology got around the rules.

Re: Nvidia contacted Anna's Archive to access books

#127

Post by bulbar »

skilled wrote: post #3

> In response, NVIDIA defended its actions as fair use, noting that books are nothing more than statistical correlations to its AI models. Does this even make sense? Are the copyright laws so bad that a statement like this would actually be in NVIDIA’s favor?

Of course it does not make sense, it's just the framing of a multi billion dollar industry and people tend to buy those.

Re: Nvidia contacted Anna's Archive to access books

#128

Post by bulbar »

thaumasiotes wrote: post #54
general1465 wrote: post #12

Earlier quoted context omitted.

Did you pirated this movie? No I did not, it is fair use because this movie is nothing more than a statistical correlation to my dopamine production.

Note that what copyright law prohibits is the action of producing a copy for someone else, not the action of obtaining a copy for yourself.

Training of an LLM however is a lossy compressing algorithm to provide a copy of a variant of the data to the user later on.

Re: Nvidia contacted Anna's Archive to access books

#129

Post by bulbar »

NitpickLawyer wrote: post #29
skilled wrote: post #3

> In response, NVIDIA defended its actions as fair use, noting that books are nothing more than statistical correlations to its AI models. Does this even make sense? Are the copyright laws so bad that a statement like this would actually be in NVIDIA’s favor?

> Does this even make sense? Are the copyright laws so bad that a statement like this would actually be in NVIDIA’s favor? It makes some sense, yeah. There's also precedent, in google scanning massive amounts of books, but not reproducing them. Most of our current copyright laws deal with reproductions. That's a no-no. It gets murky on the rest. Nvda's argument here is that they're not reproducing the works, they're…

Is they don't reproduce the data of any kind, how could the LLM be of any use?

The whole/main intention of an LLM is to reproduce knowledge.

Re: Nvidia contacted Anna's Archive to access books

#130

Post by grayhatter »

ThrowawayR2 wrote: post #34
skilled wrote: post #3

> In response, NVIDIA defended its actions as fair use, noting that books are nothing more than statistical correlations to its AI models. Does this even make sense? Are the copyright laws so bad that a statement like this would actually be in NVIDIA’s favor?

Yes, it's been discussed many times before. All the corporations training LLMs have to have done a legal analysis and concluded that it's defensible. Even one of the white papers commissioned by the FSF ( " Copyright Implications of the Use of Code Repositories to Train a Machine Learning Model " at https://www.fsf.org/licensing/copilot/copyright-implications... ), concluded that using copyrighted data to train AI wa…

> Even one of the white papers commissioned by the FSF [...] concluded that using copyrighted data to train AI was plausibly legally defensible [...] notice that the FSF has not rushed out to file copyright infringement suits even though they probably have more reason to oppose LLMs trained on FOSS code than anyone else in the world.

I agree with jkaplowitz, but for a different reason I still believe that your description feels a bit misleading to me. The FSF commissioned paper makes the argument that Microsoft's use of code FROM GITHUB, FOR COPILOT is likely non-infringing, because of the additional github ToS. This feels like critical context to provide given in the very next statement, you widened it to LLMs generally, and the FSF which likely cares about code, not on github as well.

All of that said, I'm not sure it matters, because while I don't find the argument from the that whitepaper very compelling, because it's based critically on additional grants in the ToS. IIRC (going only from memory) the ToS requires that you grant github a license as it's needed to provide the service. Github can provide the services the user reasonably understood github to provide, without violating the additional clauses specified in the existing FOSS license covering the code. That being from a while ago, and I'd say it's very murky now, because everyone knows Microsoft provides copilot, so "obviously" they need it.

Unfortunately, and importantly, when dealing with copyrights, the paper also covers the transformative fair use arguments in depth. And I do find those following arguments very compelling. The paper, (and likely others) are making the argument that the code output from an LLM is likely transformative. And thus can't be infringing compelling, (or is unlikely to be). I think in many cases, the output is clearly transformative in nature.

I've also seen code generated by claude (likely others as well?) to copy large sections from existing works. Where it's clearly "copy/paste" which clearly can't be fair use, nor transformative. The output clearly copies the soul of the work. Thus given I have no idea what dataset they're copying this code from, it's scary enough to make me unwilling to take the chance on any of it.

Post reply on HN