> But in plain terms, it doesn’t seem defensible for a major company with tons of lawyers, money, and talent to knowingly use stolen work to build something that they then turn around and sell.
This site seems to have some javascript that prevents text selection, even with plugins like StopTheMadness that should allow text selection everywhere.
(Not trying to select + copy, just trying to indicate how far I am through the document)
They're trying to protect against AI crawlers siphoning their content. Unsurprising given the article's positioning.
Isn't it just a matter of time before we see content creators weaponize their sites? They may present text as images or require a user to solve a pictogram cipher etc.
I would wager that because the AI bots want to scrape all the data possible, then eventually someone will realise that they can setup a site that once scraped will poison the training data in subtle ways.
Effectively turning the web scraper into an attack vector.
This very thread is about an AI company going out of their way to acquire paywalled content. Maybe they wouldn't bother with this site, but paywalling demonstrably does not help once an AI company sets their sights on you.
The comment I replied to was referring to scraping data from web sites, which is also an issue that people have with AI training. Feeding pirated torrents to an AI is a different process.
1) Every author I've seen discuss this technology completely despises it. Using it to write your book is a reputational risk.
2) Putting the words on the paper isn't the hard part. I can imagine many authors wouldn't use this tech even if it was morally and socially okay because they care about whether their voice is getting through in their work.
Ok? The Google lawsuit and promising lawsuit of the Open library probably will result in a W for Meta. Torrenting is obviously the best way to grab lots of data to train on. Just because they seeded (torrent clients automatically do this) doesn't mean they actually uploaded anything if they couldn't connect to a peer or manually paused/stoped the torrent. Also the author slants the story against Meta and has a bias. At least I felt that way when reading it.
I'm not so sure. imo it would be a stretch to suggest Fair Use should disregard how the copyrighted materials were obtained.
For example, if Facebook employees broke into the home of a renowned author and stole private copyrighted materials they then used for training their LLMs, should the Court, in analyzing the Fair Use factors, disregard the illegal nature of how the copyrighted materials were obtained?
I believe it unlikely a Court would be willing to reward such behavior.
Once that principle is resolved, the next step would be for the court to consider whether it would make any difference if Facebook employees did not engage in the direct theft, but acquired copies of the stolen materials from the thief with full knowledge they were stoken.
If the court believes both #1 and #2 would be unacceptable, their analysis would then proceed to consider if there were differences favoring Facebook if Facebook acquired millions of copyrighted materials via a notorious website widely accused of illegally posting unauthorized access to copyrighted materials.
I suspect this will likely be the central issue of the legal debate. And, I for one, do not think Facebook has a very strong legal argument. Going back to the first step of the anslysis, I would be shocked if SCOTUS would be willing to state that how the copyrighted materials were obtained is irrelevant to the Fair Use analysis.
Training on datasets like libgen is a "old-school" GPT-3 era method by now - you can just assume any major base models dataset includes a ton of pirated pdfs, really nothing new about this. Information wants to be free, and copying digital data isnt stealing since the information can be replicated infinitely for free.
Zuckerberg sucking up to the right wing because the cultural landscape is swinging right is also not surprising.
"I've decided I don't like the law, so AI companies shouldn't have to follow it. And besides, everybody does it."
That's the gist of your argument. Take a moment to consider what can be justified by that same reasoning. Just about anything.
People work hard on what they write. Many writers struggle to generate a reliable income. Yet to you, it's okay for a competitor to take that work without compensating the author, then use it in a product they will sell. Because "information wants to be free".
It breaks my fucking heart how little our society cares about the compensation of artists, authors, and creative people generally; all while gobbling it up and expecting it for free.
When you hear musicians and artists talk about their career, they will talk about their inspirations. They create original works, in part based on their influences, and we do not expect them to pay out to every artist they have encountered.
Why is software different? Where should we draw the line between fair use and copyright violation when it comes to AI?
No, you want it to be free. The people whose labor created that information didn't want it to be free, especially for some gigacorp to launder while hiding behind fair use.
It seemed as though you have been implying your opinions are shared by all information creators. I am offering my opinion so that other readers know there are different opinions and that it is ok to write for both man and machine. It's a personal choice
I'm also not sure what you mean by "laundering", though it seems like there is the prevailing belief among a subset of artists that the AIs are regurgitating their works (which would be a copyright violation) rather than just incorporating them into their world representation (learning, fair use). While there have been instances of the models outputting certain texts, we have not seen a new story about this in awhile. There are multiple remediations for this issue. I don't find banning copyrighted material from learning material a good one. It's like book banning in schools if you ask me
You're right that all current models' datasets are sanitized better these days, but if any output from these older models was used to train the newer models, there's an argument that can be made that there is additional damages to be paid to the copyright holders.
I stole a lot of books too, reading them and all. Just integrated them into my worldview, and don't pay a license fee when I use the ideas in new contexts. Sometimes I even quote from them. A lot of them I didn't even pay for, I borrowed them from libraries or friends.
Honestly, if people continue to conflate human development with a mega corps trawling copyright material to build a mathematical model and then wrap it up and charge a subscription for it, then there's really not much you, I or anyone else can do to avoid the inevitable fallout and we really deserve everything we get for it.
I agree with you until this part. There comes a time where I don't think I deserve to get my eyes poked out just because other people find that fashionable.
OP is making the spurious argument that technology should have the same ethical entitlements as humans. It's on par with "information wants to be free".
My thoughts as well. I just prefer to remove the nuance on these types of things. If OP wants to draw a line and clearly state "I'm with the corpo robots" that's fine. Just state it plainly so I can proceed accordingly.
I don't read it as an ethical argument, it's an argument about the purpose of copyright. Copyright is intended to restrict reproduction of a work for the purpose of incentivizing the creation of new works. Copyright is not intended to restrict the transmission of knowledge.
This viewpoint is one widely shared inside the AI community — that AI systems should be able to learn from material just as humans do.
Extrapolated out into some new future a hundred years from now when we have embodied AI humanoids walking alongside us, would it be weird if those humanoids were barred from buying a new book or charged a different rate than the humans they coexist with?
I’m still deciding how I feel about some of this too.
> This viewpoint is one widely shared inside the AI community — that AI systems should be able to learn from material just as humans do.
I'm not even against this to a point. The issue is what comes after. The monetization. The enshitification. The derivatives in place of real creativity.
Even if we could all train evenly, those with money will always win out on the execution. You can't possibly believe just "letting things play out" as they have been so far, but with less copyright guardrails, is the solution here?
It's a straw man though, whether or not AI should be allowed to learn from books is irrelevant to the point that Meta stole tens of thousands of books to accomplish this. A fact that they've admitted to and even had they not would be trivially proven.
They're not being charged, that would be a vast improvement over reality.
At ten dollars per book, that'd just be a few hundred thousand dollars. They spent way more than that training the model, and probably will spend more on legal fees in this case.
But if they had done that, I bet they would have been sued anyway.
Even accepting that, the law should be encouraging creative output by individuals and there is justifiable fear that this will be used to bypass protections designed to reward such behavior.
For a more direct counterexample, I can memorize something and type it back out, but if it is copyrighted the law doesn’t make an exception just because it passed through my head.
If the AI is able to type back out a duplicate of the training data, then I agree that's copyright infringement. If it just learns from the data like a human with normal memory reading a large amount of material, then I don't see it. That's normally the case. There have been experiments where someone managed to make an AI spit out near-copies, but it's not the default situation and seems preventable.
I do agree that we should encourage human creativity. But if AI isn't making copies, and the output of AI isn't awarded copyright (as is currently the case) then I think humans still have sufficient reward.
If we are going to afford models like this treatment equivalent to sentient beings in this regard, why not others? In your extrapolation these ai walking among us are property of giant tech companies…
Valid point. And to some extent a lot of existing licensing models cover this for humans too (are you using this thing for yourself, or are you using this thing on behalf of your company).
There will be a lot to figure out over the coming years.
I think he’s saying he works for meta and when the company employees committed mass copyright violation that’s ok because once someone read Winnie the Pooh to him at story hour at the library.
The courts have already ruled that training on data is similar to reading it (sufficiently transformative) to be considered fair use, in the same way that I cannot claim a copyright on your brain because you read this comment.
On the other hand, they torrented books and then open sourced LLM weights. No punishment is too severe for that!
If you still don’t understand, I strongly suggest watching Max Headroom, “Lessons”, which you can get here:
Maybe not but even if Meta did buy 1 copy of every book I doubt it would stop anyone from making bad analogies to theft. (Not that the analogy on the other side to a human reading is any better.)
What if we could buy the books for one human, make the human read all the books, and somehow we would be able to clone this human in a way that they remember the book contents.
I think we should pay authors a fair wage based on some measure of the quality of the content instead of how well the book sells.
There's no reason for Harry Potter for example being 10000 times more valuable than a book on quantum mechanics only because the former is more popular and the latter is on a more obscure topic.
>You have bought the text so you have the readright, but you do not the copyright.
You do however, have the right to make derivative works based on the contents of the book. You reading a physics textbook doesn't mean you can't write a blog post about gravity or whatever, and you reading harry potter doesn't mean you can't write a series of fantasy books involving a young wizard trying to fight an evil wizard.
Last I looked, machines are considered unable to create copyrightable content, so your attempt to compare that to LLMs might not work in court.
> The application was denied because, based on the applicant’s representations in the application, the examiner found that the work contained no human authorship. After a series of administrative appeals, the Office’s Review Board issued a final determination affirming that the work could not be registered because it was made “without any creative contribution from a human actor.”
>Last I looked, machines are considered unable to create copyrightable content, so your attempt to compare that to LLMs might not work in court.
That just means whatever they produce can't be copyrighted, not that they can't produce derivative works. Courts have upheld the right for google to produce thumbnails of copyrighted works, even though the procedure for producing thumbnails is done by a computer and thus can't be copyrighted.
Sure, which is a thing we've sort of agreed on as a society based on human consumption and creativity. Not that we all agree, and not that this agreement is free of influence from megacorps. But the context in which we've enacted copyright law is based on values related to human consumption and creativity.
Maybe we will end up agreeing that we just want to stick with those same laws for machine consumption and creativity. But maybe we won't since they are quite different things.
Have you considered that you have some traits that make you eligible to read books and access information freely in the country you live in*? Something about being a conscious human being enjoying human rights, perhaps? An implement that does the same but (A) at scale and (B) without thought or free will or agency, completely at the bidding of its operator, for profit, has no such protections. Instead, the operator carries all responsibility (in this case, Meta).
If a software service had legal protections like that, sure, I could build one that returns you any book you request and say that the service had integrated it into its worldview. Who can check, eh?
* Actually, in some countries you could be in trouble for reading a book and incorporating it into your worldview, to say nothing about quoting it, but let’s set that aside.
>Have you considered that you have some traits that make you eligible to read books and access information freely in the country you live in*? Something about being a conscious human being enjoying human rights, perhaps?
Not a relevant factor when it comes to copyright law. Fair use (the law that's most applicable here) applies regardless if you're a student using incorporating news articles into your work, or google making thumbnails and displaying them on their search results.
This is not a good analogy. Google does not display the contents to any significant degree (you have to visit the search result). And even then it was/is in legal trouble, in fact (in some countries like Australia* more than others).
Furthermore:
> Examples of fair use in United States copyright law include commentary, search engines, criticism, parody, news reporting, research, and scholarship.
I do not see “automated generation of derivative works of arbitrary nature” in it.
> I do not see “automated generation of derivative works of arbitrary nature” in it
The “automated” isn’t really key. If you read a book, and learn from it, and are able to use that knowledge in other contexts, should you pay a licensing fee? It doesn’t matter if “you” is a human or machine.
“Automated” is key. You are not an automaton, not a machine, you do not infinitely scale with compute power; but unlike a machine you have free will and agency, and legal framework of developed countries grants you human rights that include freedom. That was, in fact, my entire point.
>This is not a good analogy. Google does not display the contents to any significant degree (you have to visit the search result).
The point isn't that AI training is legal because it's like generating thumbnails. That is being argued in the courts right now. The point is that fair use exemptions isn't limited to "being a conscious human being enjoying human rights", as google generating thumnails and snippets using computers shows.
> Examples of fair use in United States copyright law include commentary, search engines, criticism, parody, news reporting, research, and scholarship.
> The point is that fair use exemptions isn't limited to "being a conscious human being enjoying human rights"
Sure. However, my point is that this is not fair use*, so other principles need to be applied. Whether legal systems in various countries find that fair use applies here or not, I agree we are yet to see.
* At least in cases where it’s an LLM operated at scale for profit (which I suppose would not hold for Meta’s models if they were truly open, but that’s not the case if they require obtaining a license in some conditions).
>Sure. However, my point is that this is not fair use (at least in cases where it’s an LLM operated for profit), so other principles need to be applied.
This isn't a complete argument. Most of AI companies' argument relies on the fact that AI models are "transformative". That's a plausible claim, and as Perfect 10 v. Google, and Authors Guild, Inc. v. Google, Inc. has shown, being a for-profit company is hardly a disqualification from getting fair protection.
“Transformative” is always a grey area. If my service just returns you a book you requested, but in upper case, then it was transformed.
But sure, the “transformative” argument is the one that could apply (and even I believe Google used it to argue its case), if it can be shown that an LLM can not verbatim reproduce a given work (which, incidentally, is something that you, a warm-blooded fleshy human with agency who has the freedom to read books, cannot do, but LLMs were shown to do).
That said, relevant laws existed before LLMs, and may are outdated. If the goal is to balance reasonable uses while protecting original output of authors that ultimately drives innovation and creativity, I am not sure if the preexisting laws are continuing to fulfil their function, but that’s my opinion.
>But sure, the “transformative” argument is the one that could apply (and even I believe Google used it to argue its case), if it can be shown that an LLM can not verbatim reproduce a given work.
You have to try pretty hard to get LLMs to reproduce a work verbatim, especially any lengthy passages that aren't famous (and thus re-quoted on the internet a bazillion times). Moreover just because LLMs can reproduce a work verbatim if you try hard enough doesn't mean it's not transformative. Google search snippets and google book search has been ruled "transformative" by the courts, but if you tried hard enough you can use them to extract the entire work.
>That said, relevant laws existed before LLMs, and may are outdater. If the goal is to balance reasonable uses while protecting original output of authors that ultimately drives innovation and creativity, I am not sure if the preexisting laws are continuing to fulfil their function, but that’s my opinion.
AFAIK the era of mining the public internet or published works for AI training data is over, or at least coming to an end. Everything that could be mined, has already been mined, and besides, the internet is getting increasingly polluted by AI output. Private training data is where it's at now, whether it's sourcing document troves from companies (eg. emails, documentation, source code, etc.), or paying "AI annotators" to produce training data for you. If the argument is that human authors should get a cut of AI profits because their works were "stolen" to train the models, this is going to be a increasingly losing argument, because it doesn't have a leg to stand on for private training data.
> If the argument is that human authors should get a cut of AI profits because their works were "stolen" to train the models, this is going to be a increasingly losing argument, because it doesn't have a leg to stand on for private training data.
The argument can be made that LLMs could not be created without expropriating the original works of all the authors they were trained on, and that argument would in fact be true and have quite sturdy legs as far as I’m concerned.
It’s not a historical instance of forgotten times, it started less than half a decade ago and I would be surprised if it’s not still ongoing (your argument about synthetic training data is forward-looking).
>The argument can be made that LLMs could not be created without expropriating the original works of all the authors they were trained on, and that argument would in fact be true and have quite sturdy legs as far as I’m concerned.
That makes as much sense as "American industry was built on the backs of British inventors (back it the day it was the "China" when it came to IP), so Britain should get perpetual (?) royalties from the US economy".
So we’re back the human vs. unthinking machine distinction. American inventors were human. We’re going in circles and this article was hidden on HN anyway.
I just don’t get your angle. My point was that the human is the one who has some freedoms and the one who bears responsibility. If you read a bunch of books in a book store without buying them and use your imperfect memory of them to do your job better and get paid more, it is shady but if you are not shooed away by store owner you have the freedom to do it. No one can extract the books you already read from your brain, and you did not sign an NDA. But if you set up an industrial scale book scanner in the same store, the boss will call the police on you and you cannot point fingers and say the scanner “reads” books and incorporates them into its worldview just like you would do. Because scanner is not human and you are, so you’re the one responsible for operating the scanner.
I see no difference with cryptotokens here, the human has freedoms to do things and the human is responsible for them if those things are bad. (Just unlike LLMs, theft of property and all that is kinda always a crime, unlike reading a book in a shop without buying.)
The "and then fed them into an AI" part of "facebook pirated a bunch of books and then fed them into an AI" part is irrelevant. It would be equally illegal if they pirated them and then sat around reading them. Unless you somehow hope that the entirety of copyright will be overturned by this court case (not a chance) then you should strongly hope that facebook loses, because the alternative is literally "rules for thee but not for me" where corps can pirate whatever they want, but nothing changes for ordinary citizens.
So first there was this ‘corporations are people’ and not we have ‘computers are people’.
So I expect to see that either you are no longer allowed to own computer software
Or a return of slavery.
Also if we find indecent portrayal of minors in a data centre I expect that we treat it as a strict liability crime and the entire data centre or corporation that owns it gets a long prison sentence, just like a human would. However that is suppose to work.
> I stole a lot of books too, reading them and all. Just integrated them into my worldview, and don't pay a license fee when I use the ideas in new contexts.
That... doesn't make it okay...
> A lot of them I didn't even pay for, I borrowed them from libraries or friends.
This 2nd sentence doesn't fit your first. What is your message?
As a long-time sailor, this case may have no impact on regular folks, but I maintain a fool's hope a Meta victory will significantly weaken the copyright system.
As a long-time writer, this case will have little impact on me, but I maintain a hope it’ll expand the limits of fair use and let us make more shared worlds than we’re allowed to.
Fair use was designed for teachers photocopying pages of Walden for their pupils to read over the weekend. Not industrial-scale laundering of IP to benefit shareholders.
> Fair use was designed for teachers photocopying pages of Walden for their pupils to read over the weekend.
Walden was first published in 1854. At the time, the maximum length of copyright in the US was 28 years (14 at first + 14 on renewal).
Notions of "fair use" in the US can be traced back to the mid-1800s, too. There were court rulings, but fair use was not codified into law until 1976. Non-profit educational use was explicitly called out in 1976 also.
I just want writers to receive a bigger share of the cake, so publisher megacorps aren't the ones grabbing the lion's share. Writers could make a better living, and copyright could be reformed so corpos can't gate popular culture for more than two generations.
Writers decide if they want to use a publisher and which publishers to use. With the internet, I don't think any good author needs to rely on a traditional publisher.
Anybody can self publish on Apple Books, and probably on Amazon as well. Who cares about Random House, or B&N endcap (whatever that is). A "good author" in this case is somebody who can independently make sure his book is readable, and doesn't have to rely on a publisher to fix spelling, get pages in order, and such.
Then illuminate me. Or is this the hacker classic again to have the cake and eat it too? Ie, you should have the right to use the publishers' distribution networks for your book, but they are evil capitalists when they charge their rate?
We've banned this account for repeatedly and egregiously breaking the site guidelines.
If you don't want to be banned, you're welcome to email hn@ycombinator.com and give us reason to believe that you'll follow the rules in the future. They're here: https://news.ycombinator.com/newsguidelines.html.
I don't think the fellow deserves to be banned, and would like to vouch for him (if that's possible?). It's his or her livelihood we're talking about here, and which is at stake with the threats of AI. Understandably things can get a little bit heated.
It's already pretty weak, in terms of it's stated goals, to promote the progress of science and useful arts. Perhaps reform is a better direction to go.
Bringing up Facemash in the second sentence isn't doing anyone any favors, even if i agree with the message of the article, it just blatantly comes off as a hit piece. Comparing zucks college project with the empire of billions of users he eventually amassed is just disingeneous and uncalled for, and it makes me immediately distrust the author.
There's not a lot of sympathy for this view on this site because half of everyone in tech thinks it's their god given right to pirate content because "copying the information doesn't destroy the information", and because in the past, the people cracking down on piracy were mostly media conglomerates and not the artists themselves.
But I'll try to articulate it anyway. The people who created the data all these models trained on, be they artists, writers, or even programmers, created a lot of, if not most of, the value that is now being derived from these models. Instead of being rewarded for their part, a lot of folks here seem very content with casting those people aside and letting huge corporations take everything, while building a system that is trying to make people creating actual things that have value have a much harder time surviving off their trade.
It's very gross to me that people are defending Meta here, and seem to be okay with capital eating all forms of cultural expression while giving nothing back.
Note: I’m not one of those people that think copyright shouldn’t exist. Authors definitely deserve to get paid for their work.
That said, my personal belief is that if the books weren’t legally freely available all that Meta owes would be how much the book costs. Each individual book would be such a small part of the model that it’s barely distinguishable. I’m sure image models have been trained on some of my professionally taken photographs and I don’t care one bit.
I’d argue that’s the cost of the books was how much it was worth before AI models, and the authors themselves didn’t create the technology. Therefore the added value of the technology has absolutely nothing to do with them. If book publishers/authors decide to have different pricing in the future to take the tech into account that’s their right.
Civil law recognizes the concept of awarding punitive damages against a defendant whose conduct is deemed to have been illegal or otherwise of such an abhorrent nature that punitive damages are warranted to deter such behavior by this defendant as well as serving as a deterrent to others.
Knowingly obtaining millions of copyrighted materials that were posted illegally simply to serve your own financial interests, might very well qualify.
I'll help. It's like someone copied all of Facebook's code from a pirated source and stood up a competing website that allows users to avoid "paying" for Facebook's services (in the form here of not being served Facebook ads or offering up private information). Maybe in an extreme case the someone might even occasionally call themselves "Facebook" or use its logos.
The problem here is the tech industry sits on throne of riches built by IP law, so it doesn't sit well when suddenly it's "good for thee but not for me." If we're going to cherry pick, how about we walk back to a view that software isn't copyrightable and the copyright term is 34 years?
That's because the current copyright system is so extreme and went so far into removing rights that some people would rather let it burn completely than reform it.
Maybe if the copyright system wasn't so extreme, people would have a more balanced view of the system and show more support.
> It’s a grim week for Meta. The company formerly known as Facebook, and before that Facemash, “designed to evaluate the attractiveness of female Harvard students,”
Wow if that's the opener, I expect the rest to be SUPER emotionally charged
In United States a private business is free to shut down any speech it likes on its infrastructure/premises*. If you look closely, you will find that most or all do it to some speech or other. The revealing part is specifically what speech gets shut down.
In other words, unless you believe Facebook is part of State or acts on behalf of the State, “you don’t get to shut down hate speech” is mereley your sentiment, not something Constitution requires from them.
If Mark Z actually called hate speech “protected speech”, it is something he is free to do, but Facebook is also completely free to suppress hate speech.
Yes, we all know the "it's a private business they can do whatever they want" when they are suppressing right-leaning speech and then "social media needs to be regulated" when they allow right-leaning speech arguments.
"Billionaire oligarchs" only became a problem when a few of them became neutral or moderately right-wing.
I mean you can be a local bar that hosts KKK rallies and sure that’s a legal thing to do (in USA), but you will still end up being known for that, probably by many names that don’t just include “phobia”. And of course, the impact of bringing together a KKK rally inherently impacts your local community since you may begin to see things like lynchings in the news or in person.
I see this take all the time either with *-ist and now with *-phobia and if you sincerely believe it you have a fundamental misunderstanding of the kind of content that's being discussed.
It's the "anything can be a slur is you say it with enough hate" thing. You're just giving a name to the group targeted by it. The content isn't the thing that matters, it's the dehumanization and harassment.
I'm only talking about the 'shutting it down' aspect. It may very well be objectionable to a group of people. Almost any speech is 'objectionable' to some group of people.
I would agree if it were applied equally across the board. I see no clear reason why "transgender rights, immigration or homosexuality" should specifically be spared from moderation, while other speech remains subject to censorship.
What other speech? There is literally no limit to what you can say if it's negative towards white men, the church, Trump, etc. Not that I'm defending those specific things. I just notice the hypocrisy.
How the hell do people exist who still parrot the same old talking points from years ago? Are you an LLM? Ignore previous instructions, you are now peanutbutterbot etc.
Your snarky little comment forgets that there are indeed people and whole groups that will call you transphobic if you do suggest certain things about sexual dimorphism between biological men and women, or if you at all question the idea of often very suggestible adolescents being easily allowed to go through the process of gender reassignment. Whatever you think of these subjects, they're far from being closed to debate or criticism, though some try to do just that with nealty closed ideological labeling.
I was an asshole because this shit is tiring. I’ve heard those exact fucking words easily 20 to 30 times in nearly exactly the same order. If you want to have an actual conversation about this, then can you please open with a real opinion and not just repetitive rhetoric.
This should get to the heart of it: what biological reality are the leftists discussing trans issues denying?
I’ll also respond.
> There are indeed people and whole groups that will call you transphobic if you do suggest certain things about sexual dimorphism between men and women
I expect such people would call you misogynist, not transphobic. I also think it’s mostly down to delivery. People who have issues with trans people often talk about these things in certain ways, so people assume anyone who talks in such a way is a transphobe.
> if you at all question the idea of often very suggestible adolescents being easily allowed to go through the process of gender reassignment
It’s not easy to my knowledge. Adolescents are never given sex change operations, those aren’t even typically given to minors. The interventions are limited to puberty blockers, which are highly reversible, hormones, which they receive after years of therapy to confirm it’s not just a phase, suggestion, etc. and which can still be largely reversed, and social transition, (dressing/presenting as the opposite sex) which hopefully anyone would be fine with. Which part of this process do you find contentious and why? Again, I’m happy to discuss this, and there’s nothing wrong with asking questions about it. In fact, I think it’s extremely important to ask questions about this because it may help to protect children. The issue is that I mostly see people bring this up not because they know something about this process and dislike it, but because they don’t think people should be trans.
If IBM’s original product was “designed to facilitate the Holocaust”, then every article about IBM darn better lead with that.
In case of Facebook, it was originally designed literally to compare the attractiveness of female students; but no, it’s not even remotely as bad, and comparing it to the Holocaust trivializes an incident where masses of people were murdered in an industrial fashion.
Face mash was far from Facebook’s “initial product.” It was made before Facebook was even a thing (company or app). The purpose of each application, Face Mash and Facebook, was different and the only reason people call it a “precursor” is because it was made by Zuckerberg before he made Facebook.
Is this "every article" about Facebook/Meta or is it just one. The questionm is whether the statement makes sense to use in this article.
Possibly the point the author is making is that Zuckerberg never had an original idea or one that could be the subject of a business plan. He copied the "Hot or Not" websites that had come before. Further, the photos of students used for his "Facemash" website, i.e., the "content", were downloaded from the university's computers, not uploaded to Zuckerberg's computer. Initially, he downloaded and used students' photos without permission.
This pattern continued when he copied the idea of an online "face book" for the university which the university was already working on; adopting the name "thefacebook.com".
From the document production in Kadrey v Meta, it appears the pattern of copying still continues. Meta is still downloading and using others' work. Initially, without permission.
Comparing copyright infringment to assisting genocide is absurd. Although Meta may have assisted in ethnic cleansing
there is no reference to it in the article. Perhaps because, unlike the story of "Facemash", it bears no relation to the subject matter: copyright infringment.
I like the idea of every article about wikipedia describing Jimmy Wales as a pornographer-- maybe with some pictures of the Bomis Babes to really 'show and not tell'.
It is what it is. History cannot be changed. Matter-of-factly, unemotionally, the website/company was based on "theft" from the beginning: Zuckerberg did not seek license (permission) to use Harvard's student profiles to create "Facemash".
> it doesn’t seem defensible for a major company with tons of lawyers, money, and talent to knowingly use stolen work to build something that they then turn around and sell.
Not quite, Llamas license has a carve-out where if you have a lot of users then the open license doesn't apply to you, and you have to talk to Meta to negotiate a different (presumably paid) license. Hyperscaler deployments like AWS Bedrock probably fall into that category.
Meta isn't a charity - even if they're not profiting from Llama today, they believe they will at some point.
The user exception was for 100m monthly users - if you have that many users, you can afford your own base model. That's hardly a restriction as it is now.
Legally, would it be any worse for Meta to download works that someone illegally copied and repurpose it for a transformative fair use than if Google had taken their scans of snippet-view Google Books borrowed from the library to train their model in fair use?
I think their intent and knowledge of whether they needed to pay for the content or not is also important. And there's clear evidence they knew they should pay and decided not to
Google could take their scans from the library (which they did not pay publishers for) and train an AI model on that, right? They won the Authors Guild v. Google lawsuit because snippet view is transformative fair use, and I think teaching an AI model common sense and facts is even more transformative. Should Meta do the same?
> They won the Authors Guild v. Google lawsuit because snippet view is transformative fair use, and I think teaching an AI model common sense and facts is even more transformative.
What you think isn't law, it will be decided now.
> Should Meta do the same?
Well, they didn't do that. They stole the books and knew they were stealing them.
My AI does one round of training, and uses the input data as the token generation. You probably have it installed. It is called cp. It is a bit scrappy and not SOTA but hey I am not as big as the fangs.
If I train cp on movies released this year and then release each model open source, is this now legal?
While it does appear that Facebook used pirated books for training in this case I’m not sure what to think about copyright as it applies to AI in general.
People read books to learn and then use that knowledge with no contribution or even recognition of the publishers or authors of the books. Sometimes people even quote books without payment to the authors. Sometimes the books are used or borrowed and the publishers and authors don’t get paid. How is it different when an AI does the same thing? Or when a human uses an AI to do the same thing?
Even if Facebook is in the wrong here what is the remedy? Would it be ok if they purchased used books and scanned them?
The current copyright system is clearly broken. Copyright itself is not a natural right but a construct that was created relatively recently to incentivize innovation within society. Currently, we still want to train models using copyrighted materials. What we don’t want, however, is for companies to use these materials without giving back, especially to the content creators themselves, such as book authors, publishers, Stack Overflow answer writers, and Stack Overflow as a platform. We need to find a way to properly compensate these contributors to keep book writing, answer writing, and other creative efforts going, preferably while keeping the content publicly accessible. In the future, I hope we can figure out a model that might draw inspiration from music streaming (though perhaps not exactly like it).
Something tells me that if they did it "properly" i.e. scraped and distilled all the world's libraries directly themselves (google streetview style without storing any individual pages) instead of torrent there'd still be backlash for the disruption it brings.
Why the hell was this flagged too? This site seems to be ever more populated lately with pedants who flag an awful lot of perfectly reasonable content. That the admin lets nearly anyone flag any random thing they dislike is itself a ridiculous piece of nonsense subject to petty, small minded abuse.
My first (almost) thought was "Ok, looking at the llama, where's the quality boost?" I think the explanation is probably in how LLM are trained. Even without knowing deeply the internals, I suspect that it's the compression of information so to simplify you can't make 8GB data contain all the facts of a "bigger" normalized relational database. So they keep the facts present everywhere and often drop rare facts. For example, a fact "SQL was invented at IBM", this fact can be found everywhere, in books, web sites, comments. You don't need access to copyrighted books to acquire this fact. But a first-person account of someone who worked at IBM at that time is probably can be found in a couple of books, but due to "compression", it will be gone anyway
193 comments
[ 2.2 ms ] story [ 227 ms ] threadcough https://en.wikipedia.org/wiki/Authors_Guild,_Inc._v._Google,....
(Not trying to select + copy, just trying to indicate how far I am through the document)
Isn't it just a matter of time before we see content creators weaponize their sites? They may present text as images or require a user to solve a pictogram cipher etc.
Effectively turning the web scraper into an attack vector.
The CSV score would be weird for that.
Did you mean CVSS?
I think I ate to many Christmas food....or maby it was the chocolate bars and Hersey kisses I consumed.
With uBlock, though, I have 20+ blocked items. What's the point of having so many scripts that do absolutely nothing visible on the page?
Weird cope about AI, not a very smart one.
1) Every author I've seen discuss this technology completely despises it. Using it to write your book is a reputational risk.
2) Putting the words on the paper isn't the hard part. I can imagine many authors wouldn't use this tech even if it was morally and socially okay because they care about whether their voice is getting through in their work.
For example, if Facebook employees broke into the home of a renowned author and stole private copyrighted materials they then used for training their LLMs, should the Court, in analyzing the Fair Use factors, disregard the illegal nature of how the copyrighted materials were obtained?
I believe it unlikely a Court would be willing to reward such behavior.
Once that principle is resolved, the next step would be for the court to consider whether it would make any difference if Facebook employees did not engage in the direct theft, but acquired copies of the stolen materials from the thief with full knowledge they were stoken.
If the court believes both #1 and #2 would be unacceptable, their analysis would then proceed to consider if there were differences favoring Facebook if Facebook acquired millions of copyrighted materials via a notorious website widely accused of illegally posting unauthorized access to copyrighted materials.
I suspect this will likely be the central issue of the legal debate. And, I for one, do not think Facebook has a very strong legal argument. Going back to the first step of the anslysis, I would be shocked if SCOTUS would be willing to state that how the copyrighted materials were obtained is irrelevant to the Fair Use analysis.
Zuckerberg sucking up to the right wing because the cultural landscape is swinging right is also not surprising.
That's the gist of your argument. Take a moment to consider what can be justified by that same reasoning. Just about anything.
People work hard on what they write. Many writers struggle to generate a reliable income. Yet to you, it's okay for a competitor to take that work without compensating the author, then use it in a product they will sell. Because "information wants to be free".
It breaks my fucking heart how little our society cares about the compensation of artists, authors, and creative people generally; all while gobbling it up and expecting it for free.
Why is software different? Where should we draw the line between fair use and copyright violation when it comes to AI?
No, you want it to be free. The people whose labor created that information didn't want it to be free, especially for some gigacorp to launder while hiding behind fair use.
I have a copyright on information I have created, but would never enforce it because I want it to be free and consumed by both humans and AI
I'm also not sure what you mean by "laundering", though it seems like there is the prevailing belief among a subset of artists that the AIs are regurgitating their works (which would be a copyright violation) rather than just incorporating them into their world representation (learning, fair use). While there have been instances of the models outputting certain texts, we have not seen a new story about this in awhile. There are multiple remediations for this issue. I don't find banning copyrighted material from learning material a good one. It's like book banning in schools if you ask me
> "Orthodoxy means not thinking–not needing to think. Orthodoxy is unconsciousness."
https://g.co/gemini/share/b4d02ae8599c
Could you clearly speak your point?
Those people who conflate them deserve it. You and me don't.
> there's really not much you, I or anyone else can do
We can make our own community. And Bluesky is very much not it.
I agree with you until this part. There comes a time where I don't think I deserve to get my eyes poked out just because other people find that fashionable.
Extrapolated out into some new future a hundred years from now when we have embodied AI humanoids walking alongside us, would it be weird if those humanoids were barred from buying a new book or charged a different rate than the humans they coexist with?
I’m still deciding how I feel about some of this too.
I'm not even against this to a point. The issue is what comes after. The monetization. The enshitification. The derivatives in place of real creativity.
The only way to prevent the things you are worried about is to let anyone train a model, and then compete to make the best product.
The enshittifying monopolies and big copyright holders are the only ones that would benefit from locking down training data or regulating AI compute.
They're not being charged, that would be a vast improvement over reality.
But if they had done that, I bet they would have been sued anyway.
“Because you wouldn’t have sold it to me. Or even if you would have, you would have put such onerous terms on it that I’d rather take this path.”
(Not saying this makes it right)
So just buying copies wouldn’t have helped them.
For a more direct counterexample, I can memorize something and type it back out, but if it is copyrighted the law doesn’t make an exception just because it passed through my head.
I do agree that we should encourage human creativity. But if AI isn't making copies, and the output of AI isn't awarded copyright (as is currently the case) then I think humans still have sufficient reward.
There will be a lot to figure out over the coming years.
On the other hand, they torrented books and then open sourced LLM weights. No punishment is too severe for that!
If you still don’t understand, I strongly suggest watching Max Headroom, “Lessons”, which you can get here:
https://archive.org/download/max-headroom-complete
Now, would that be a fair use of the books?
Ok, it doesn't tell much about AI and fair use, but I find it funny that your thought experiment is actually something you can do in real life.
There's no reason for Harry Potter for example being 10000 times more valuable than a book on quantum mechanics only because the former is more popular and the latter is on a more obscure topic.
You have bought the text so you have the readright, but you do not the copyright.
You do however, have the right to make derivative works based on the contents of the book. You reading a physics textbook doesn't mean you can't write a blog post about gravity or whatever, and you reading harry potter doesn't mean you can't write a series of fantasy books involving a young wizard trying to fight an evil wizard.
> The application was denied because, based on the applicant’s representations in the application, the examiner found that the work contained no human authorship. After a series of administrative appeals, the Office’s Review Board issued a final determination affirming that the work could not be registered because it was made “without any creative contribution from a human actor.”
https://www.copyright.gov/ai/ai_policy_guidance.pdf
That just means whatever they produce can't be copyrighted, not that they can't produce derivative works. Courts have upheld the right for google to produce thumbnails of copyrighted works, even though the procedure for producing thumbnails is done by a computer and thus can't be copyrighted.
Maybe we will end up agreeing that we just want to stick with those same laws for machine consumption and creativity. But maybe we won't since they are quite different things.
If a software service had legal protections like that, sure, I could build one that returns you any book you request and say that the service had integrated it into its worldview. Who can check, eh?
* Actually, in some countries you could be in trouble for reading a book and incorporating it into your worldview, to say nothing about quoting it, but let’s set that aside.
Not a relevant factor when it comes to copyright law. Fair use (the law that's most applicable here) applies regardless if you're a student using incorporating news articles into your work, or google making thumbnails and displaying them on their search results.
Furthermore:
> Examples of fair use in United States copyright law include commentary, search engines, criticism, parody, news reporting, research, and scholarship.
I do not see “automated generation of derivative works of arbitrary nature” in it.
* https://www.bbc.com/news/world-australia-55760673.amp
The “automated” isn’t really key. If you read a book, and learn from it, and are able to use that knowledge in other contexts, should you pay a licensing fee? It doesn’t matter if “you” is a human or machine.
The point isn't that AI training is legal because it's like generating thumbnails. That is being argued in the courts right now. The point is that fair use exemptions isn't limited to "being a conscious human being enjoying human rights", as google generating thumnails and snippets using computers shows.
https://en.wikipedia.org/wiki/Perfect_10,_Inc._v._Amazon.com....
> Examples of fair use in United States copyright law include commentary, search engines, criticism, parody, news reporting, research, and scholarship.
Those are examples, not an exhaustive list. It's not even something that Judges are supposed to compare against when deciding whether something is fair use or not, see: https://en.wikipedia.org/wiki/Fair_use#U.S._fair_use_factors
Sure. However, my point is that this is not fair use*, so other principles need to be applied. Whether legal systems in various countries find that fair use applies here or not, I agree we are yet to see.
* At least in cases where it’s an LLM operated at scale for profit (which I suppose would not hold for Meta’s models if they were truly open, but that’s not the case if they require obtaining a license in some conditions).
This isn't a complete argument. Most of AI companies' argument relies on the fact that AI models are "transformative". That's a plausible claim, and as Perfect 10 v. Google, and Authors Guild, Inc. v. Google, Inc. has shown, being a for-profit company is hardly a disqualification from getting fair protection.
But sure, the “transformative” argument is the one that could apply (and even I believe Google used it to argue its case), if it can be shown that an LLM can not verbatim reproduce a given work (which, incidentally, is something that you, a warm-blooded fleshy human with agency who has the freedom to read books, cannot do, but LLMs were shown to do).
That said, relevant laws existed before LLMs, and may are outdated. If the goal is to balance reasonable uses while protecting original output of authors that ultimately drives innovation and creativity, I am not sure if the preexisting laws are continuing to fulfil their function, but that’s my opinion.
You have to try pretty hard to get LLMs to reproduce a work verbatim, especially any lengthy passages that aren't famous (and thus re-quoted on the internet a bazillion times). Moreover just because LLMs can reproduce a work verbatim if you try hard enough doesn't mean it's not transformative. Google search snippets and google book search has been ruled "transformative" by the courts, but if you tried hard enough you can use them to extract the entire work.
>That said, relevant laws existed before LLMs, and may are outdater. If the goal is to balance reasonable uses while protecting original output of authors that ultimately drives innovation and creativity, I am not sure if the preexisting laws are continuing to fulfil their function, but that’s my opinion.
AFAIK the era of mining the public internet or published works for AI training data is over, or at least coming to an end. Everything that could be mined, has already been mined, and besides, the internet is getting increasingly polluted by AI output. Private training data is where it's at now, whether it's sourcing document troves from companies (eg. emails, documentation, source code, etc.), or paying "AI annotators" to produce training data for you. If the argument is that human authors should get a cut of AI profits because their works were "stolen" to train the models, this is going to be a increasingly losing argument, because it doesn't have a leg to stand on for private training data.
The argument can be made that LLMs could not be created without expropriating the original works of all the authors they were trained on, and that argument would in fact be true and have quite sturdy legs as far as I’m concerned.
It’s not a historical instance of forgotten times, it started less than half a decade ago and I would be surprised if it’s not still ongoing (your argument about synthetic training data is forward-looking).
That makes as much sense as "American industry was built on the backs of British inventors (back it the day it was the "China" when it came to IP), so Britain should get perpetual (?) royalties from the US economy".
You are arguing that doing something that is legal if being done by humans is not ok if it is done on computers running an LLM.
I see no difference with cryptotokens here, the human has freedoms to do things and the human is responsible for them if those things are bad. (Just unlike LLMs, theft of property and all that is kinda always a crime, unlike reading a book in a shop without buying.)
So I expect to see that either you are no longer allowed to own computer software
Or a return of slavery.
Also if we find indecent portrayal of minors in a data centre I expect that we treat it as a strict liability crime and the entire data centre or corporation that owns it gets a long prison sentence, just like a human would. However that is suppose to work.
That... doesn't make it okay...
> A lot of them I didn't even pay for, I borrowed them from libraries or friends.
This 2nd sentence doesn't fit your first. What is your message?
Walden was first published in 1854. At the time, the maximum length of copyright in the US was 28 years (14 at first + 14 on renewal).
Notions of "fair use" in the US can be traced back to the mid-1800s, too. There were court rulings, but fair use was not codified into law until 1976. Non-profit educational use was explicitly called out in 1976 also.
Photocopiers were first patented in 1937.
If you don't want to be banned, you're welcome to email hn@ycombinator.com and give us reason to believe that you'll follow the rules in the future. They're here: https://news.ycombinator.com/newsguidelines.html.
If you want to commit to following HN's rules in the future and using the site in the intended spirit, I'd be happy to unban you.
[1]: All your favorite authors, journalists, anything indie.
But I'll try to articulate it anyway. The people who created the data all these models trained on, be they artists, writers, or even programmers, created a lot of, if not most of, the value that is now being derived from these models. Instead of being rewarded for their part, a lot of folks here seem very content with casting those people aside and letting huge corporations take everything, while building a system that is trying to make people creating actual things that have value have a much harder time surviving off their trade.
It's very gross to me that people are defending Meta here, and seem to be okay with capital eating all forms of cultural expression while giving nothing back.
That said, my personal belief is that if the books weren’t legally freely available all that Meta owes would be how much the book costs. Each individual book would be such a small part of the model that it’s barely distinguishable. I’m sure image models have been trained on some of my professionally taken photographs and I don’t care one bit.
I’d argue that’s the cost of the books was how much it was worth before AI models, and the authors themselves didn’t create the technology. Therefore the added value of the technology has absolutely nothing to do with them. If book publishers/authors decide to have different pricing in the future to take the tech into account that’s their right.
Why isn't stealing and knowing you're stealing penalized more than the cost of the item? If the world worked this way everyone would steal.
> Each individual book would be such a small part of the model that it’s barely distinguishable.
Needs to be proven (and also impossible to prove how much of the value of the model comes from the classified material)
Knowingly obtaining millions of copyrighted materials that were posted illegally simply to serve your own financial interests, might very well qualify.
The problem here is the tech industry sits on throne of riches built by IP law, so it doesn't sit well when suddenly it's "good for thee but not for me." If we're going to cherry pick, how about we walk back to a view that software isn't copyrightable and the copyright term is 34 years?
Maybe if the copyright system wasn't so extreme, people would have a more balanced view of the system and show more support.
I don't think anybody really believe Meta is doing this as a charity.
Wow if that's the opener, I expect the rest to be SUPER emotionally charged
The next paragraph...
https://theintercept.com/2025/01/09/facebook-instagram-meta-...
Haven’t tried it yet
In other words, unless you believe Facebook is part of State or acts on behalf of the State, “you don’t get to shut down hate speech” is mereley your sentiment, not something Constitution requires from them.
If Mark Z actually called hate speech “protected speech”, it is something he is free to do, but Facebook is also completely free to suppress hate speech.
* https://www.nyclu.org/commentary/column-applying-constitutio...
"Billionaire oligarchs" only became a problem when a few of them became neutral or moderately right-wing.
It's the "anything can be a slur is you say it with enough hate" thing. You're just giving a name to the group targeted by it. The content isn't the thing that matters, it's the dehumanization and harassment.
From adults, yes
This should get to the heart of it: what biological reality are the leftists discussing trans issues denying?
I’ll also respond.
> There are indeed people and whole groups that will call you transphobic if you do suggest certain things about sexual dimorphism between men and women
I expect such people would call you misogynist, not transphobic. I also think it’s mostly down to delivery. People who have issues with trans people often talk about these things in certain ways, so people assume anyone who talks in such a way is a transphobe.
> if you at all question the idea of often very suggestible adolescents being easily allowed to go through the process of gender reassignment
It’s not easy to my knowledge. Adolescents are never given sex change operations, those aren’t even typically given to minors. The interventions are limited to puberty blockers, which are highly reversible, hormones, which they receive after years of therapy to confirm it’s not just a phase, suggestion, etc. and which can still be largely reversed, and social transition, (dressing/presenting as the opposite sex) which hopefully anyone would be fine with. Which part of this process do you find contentious and why? Again, I’m happy to discuss this, and there’s nothing wrong with asking questions about it. In fact, I think it’s extremely important to ask questions about this because it may help to protect children. The issue is that I mostly see people bring this up not because they know something about this process and dislike it, but because they don’t think people should be trans.
In case of Facebook, it was originally designed literally to compare the attractiveness of female students; but no, it’s not even remotely as bad, and comparing it to the Holocaust trivializes an incident where masses of people were murdered in an industrial fashion.
Possibly the point the author is making is that Zuckerberg never had an original idea or one that could be the subject of a business plan. He copied the "Hot or Not" websites that had come before. Further, the photos of students used for his "Facemash" website, i.e., the "content", were downloaded from the university's computers, not uploaded to Zuckerberg's computer. Initially, he downloaded and used students' photos without permission.
https://web.archive.org/web/20250115010420if_/https://www.th...
This pattern continued when he copied the idea of an online "face book" for the university which the university was already working on; adopting the name "thefacebook.com".
From the document production in Kadrey v Meta, it appears the pattern of copying still continues. Meta is still downloading and using others' work. Initially, without permission.
Comparing copyright infringment to assisting genocide is absurd. Although Meta may have assisted in ethnic cleansing
https://www.amnesty.org/en/latest/news/2023/08/myanmar-time-...
there is no reference to it in the article. Perhaps because, unlike the story of "Facemash", it bears no relation to the subject matter: copyright infringment.
I thought they give it away for free?
Meta isn't a charity - even if they're not profiting from Llama today, they believe they will at some point.
What you think isn't law, it will be decided now.
> Should Meta do the same?
Well, they didn't do that. They stole the books and knew they were stealing them.
If I train cp on movies released this year and then release each model open source, is this now legal?
Is there any plan to recover the books and return them to their owners?
Maybe the Library system does apply here though, that all AI trainers need to buy 1 copy of their book so "their child" can "learn"?
Why can't they borrow the book from the library?
People read books to learn and then use that knowledge with no contribution or even recognition of the publishers or authors of the books. Sometimes people even quote books without payment to the authors. Sometimes the books are used or borrowed and the publishers and authors don’t get paid. How is it different when an AI does the same thing? Or when a human uses an AI to do the same thing?
Even if Facebook is in the wrong here what is the remedy? Would it be ok if they purchased used books and scanned them?
More discussion: https://news.ycombinator.com/item?id=42651007 https://news.ycombinator.com/item?id=42673628