Generative artificial intelligence (GenAI) company Anthropic has claimed to a US court that using copyrighted content in large language model (LLM) training data counts as “fair use”, however.
Under US law, “fair use” permits the limited use of copyrighted material without permission, for purposes such as criticism, news reporting, teaching, and research.
In October 2023, a host of music publishers including Concord, Universal Music Group and ABKCO initiated legal action against the Amazon- and Google-backed generative AI firm Anthropic, demanding potentially millions in damages for the allegedly “systematic and widespread infringement of their copyrighted song lyrics”.
…then maybe they shouldn’t exist. If you can’t pay the copyright holders what they’re owed for the license to use their materials for commercial use, then you can’t use ‘em that way without repercussions. Ask any YouTuber.
You might want to read this article by Kit Walsh, a senior staff attorney at the EFF, and this one by Katherine Klosek, the director of information policy and federal relations at the Association of Research Libraries. YouTube’s one-sided strike-happy system isn’t the real world.
Headlines like these let people assume that it’s illegal, rather than educate them on their rights.
When Annas-Archive or Sci-Hub get treated the same as these giant corporations, I’ll start giving a shit about the “fair use” argument.
When people pirate to better the world by increasing access to information, the whole world gets together to try to kick them off the internet.
When giant companies with enough money to make Solomon blush pirate to make more oodles of money and not improve access to information, it’s “fAiR uSe.”
Literally everyone knew from the start that books3 was all pirated and from ebooks with the DRM circumvented and removed. It was noted when it was created it was basically the entirety of private torrent tracker Bibliotik.
AI training should not be a privilege of the mega-corporations. We already have the ability to train open source models, and organizations like Mozilla and LAION are working to make AI accessible to everyone. We can’t allow the ultra-wealthy to monopolize a public technology by creating barriers that make it prohibitively expensive for regular people to keep up. Mega corporations already have a leg up with their own datasets and predatory terms of service that exploit our data. Don’t do their dirty work for them.
Denying regular people access to a competitive, corporate-independent tool for creativity, education, entertainment, and social mobility, we condemn them to a far worse future, with fewer rights than we started with.
How am I doing their dirty work for them? I literally will stop thinking that they’re getting away with piracy for profit when we stop haranguing people who are committing to piracy for the benefit of mankind.
I’m not saying Meta should be stopped, I’m saying the prosecution of Sci-Hub and Annas-Archive need to be stopped under the same pretenses.
If it’s okay to pirate for the purpose of making money (what we put The Pirate Bay admins in jail for), then it’s okay to pirate to benefit mankind.
There is literally no way in hell someone can convince me what Meta and others are doing is not pirating to use the data contained within to make money. What’s good for the goose is good for the gander, as they say.
I reiterate, they knew it was pirated and had DRM circumvented when they downloaded it. There was zero question of the source of this data. They knew from the beginning they intended to profit from the use of this data. How is that different than what we accused The Pirate Bay admins of?
It really feels like “Well these corporations have money to steal more prolifically than little people, so since they’re stealing is so big, we have to ignore it.”
Then I misunderstood what you were saying. Carry on.
You don’t see the difference between distributing someone else’s content against their will and using their content for statistical analysis? There’s a pretty clear difference between the two, especially as fair use is concerned.
By and large copyright infringement is illegal. That some things aren’t infringement doesn’t change that a general stance of “if I don’t have permission, I can’t copy it” is correct. The first argument in the EFF article is effectively the title: “it can’t be copyright, because otherwise massive AI models would be impossible to build”. That doesn’t make it fair use, they just want it to become so.
deleted by creator
You do realize that there may in fact be different, distinct groups of Lemmy users with differing, potentially non-overlapping beliefs, yeah?
deleted by creator
Oh, well, you’ve clearly done the kind of deep and thoughtful analysis that would allow you to determine the general opinions of all Lemmy users. My mistake. Carry on.
deleted by creator
Using copyrighted material for something you aren’t gonna make any money off of? Cool, go hog wild. If you’re gonna use some music or art that you didn’t make in something that will make you money, the folks that made whatever you used should get a cut. Not the whole cut, but a cut.
If an artist falls in love with drawing and learns to draw from Jack Kirby’s work and at the beginning even imitates his style, does he owe Jack Kirby royalties for every drawing he does as he ‘learned’ on Jack’s copyrighted art?
I think in that case, no. ‘Style’ is one thing, directly using someone’s art in your own work is something else entirely. However, we’re talking about a person here, not a program developed by a company for the express purpose of making as much money as possible in the shortest amount of time. Until AI can truly demonstrate that it is truly thinking and not simply executing commands given, I don’t think the lines are blurred nearly enough to suggest that someone learning to paint and an AI trained on hundreds of thousands of pieces of art for the purpose of making money for the company that built it are remotely the same.
deleted by creator
In what way? I rephrased my original comment.
And corporations want people to pay for it but they don’t want to pay for it themselves. It’s almost as if no one likes copyright, but it benefits some ppl more than others.
You do realize that lemmy contains very many users, many of whom disagree on any number of things. You are randomly assigning the opinions of lemmy’s pirate users to a random commenter without evidence that they actually hold those opinions, because it’d be convenient for you if they’re contradicting themself in any way (though the degree to which that would be a contradiction is also arguable). It’s just a way of constructing a strawman instead of engaging with your interlocutor’s actual words.
Also, part of the problem is that these LLMs very often do directly copy and spit out articles and random forum posts and etc word-for-word verbatim, or it’ll do something that’s the equivalent of a plagiarist who swaps a few words around in a sad attempt to not get caught. It becomes especially likely depending on how specific the search is, like if you look for a niche topic hardly anyone has written extensively on or for the solution to an esoteric problem that maybe just one person on a forum somewhere found an answer to. It also typically does not even give credit or link to its sources.
Plus, copyright law, if it exists, must apply to everyone, including major coporations. That’s a separate issue than whether or not copyright law needs reform (it obviously does). If you wanna abolish copyright, fine, ok, get it abolished through the government. But while copyright law is still the law, I’m not ozk with giving magacorps a pass to break it legally, especially when they’re more than happy to sue random, harmless individuals for violating their own copyrights. They want the law not to apply to them because they’re rich.
The argument they’re making is just ridiculous on its face when you compare it to other crimes. If AI should be allowed to violate copyright because otherwise it can’t exist as it is, then anyone should be able to violate copyright because otherwise their cool projects won’t be able to exist. And I should be able to rob a bank because otherwise I won’t have all that money. You should be able to commit murder because otherwise your annoying coworker will keep bugging you. She should be able to walk out of a store with an iPhone without paying for it because otherwise she won’t have an iPhone. Etc. It’s an argument that says the criminal’s motivations are legal justification for the crime. “You should let me legally do the thing because otherwise I can’t do the thing” is just not a convincing argument in my book.
deleted by creator
Then it shouldn’t exist.
This isn’t an issue of fair use. They’re stealing other people’s work and using it to create something new and then trying to profit from it, without any credit or recompense.
Just like I do with literally all content I’ve ever consumed. Everything I’ve seen has been remashed in my brain into the competencies I charge money for.
It’s not until I profit off of someone else’s work — ie when the source of the profit is their work — that I’m breaking any rules.
This is a non-issue. We’ve let our (legitimate) fear of AI twist us into distorting truth and motivated reasoning. Instead of trying to paint AI as morally wrong, we should admit that we are afraid of it.
We’re trying to replace our fear with disgust and anger. It’s not healthy for us. AI is ultra fucking scary. And not because it’s going to take inspiration from a copyrighted song when it writes a different song. AI is ultra fucking scary because it will soon surpass any possibility of our understanding, and we will be at the whim of an alien intelligence.
But that’s too sci fi sounding, to be something people have to look at. Because it sounds so out there, it’s easy to scoff at and dismiss. So instead of acknowledging our terror at the fact this thing will likely end humanity as we know it, we’re sublimating that energy through righteous indignation. See, indignation is unpleasant, but it’s less threatening to the self than terror.
It’s understandable, like doing another line of coke is understandable. But it is not healthy, not productive, and will not play out the way we think. We need to stop letting our fear turn our minds to mush.
Reading someone else’s material before you write new material is not the same as copying someone else’s material and selling it as your own. The information on the internet has always been considered free for legal use. And the limit of legal use is based on the selling of others’ verbatim material.
This is a simple fact, easy to see. Except recognizing it nullifies the righteous indignation, opening the way for the terror and confusion to come in again.
Now that it exists how do you propose we make it not exist?
Even if we outlaw it Russia and China won’t and without the tools to fight back against it the web is basically done as anything but a propaganda platform
It doesn’t matter what business we’re talking about. If you can’t afford to pay the costs associated with running it, it’s not a viable business. It’s pretty fucking simple math.
And no, we’re not talking about “to big to fail” business (that SHOULD be allowed to fail, IMHO) we’re talking about AI, that thing they keep trying to shove down our throats and that we keep saying we don’t want or need.
Why are people publishing so much content online if they aren’t cool with people downloading it? Like, the web is an open platform. The content is there for the taking.
Until one of these AIs just starts selling other people’s work as its own, and no I don’t mean derivative work I mean the copyrighted material, nobody is breaking the rules here.
I read content online without paying for a license. I should only have to obtain a license for material I’m publishing, not material I read.
Until one of these AIs just starts selling other people’s work as its own, and no I don’t mean derivative work I mean the copyrighted material, nobody is breaking the rules here.
Except of course that’s not how copyright law works in general.
Of course the questions are 1) is training a model fair use and 2) are the resulting outputs derivative works. That’s for the courts to decide.
But in general, just because I publish content on my website, does not give anyone else license or permission to republish that content or create derivative works, whether for free or for profit, unless I explicitly license that content accordingly.
That’s why things like Creative Commons exists.
But surely you already knew that.
Right, but I think it’s going to be a tough legal argument that using a text to adjust database weighting links between word associations is copying or distributing any part of that work. Assuming courts understand the math/algorithms.
deleted by creator
I don’t know if you noticed this but some really big companies with high stock valuations are only existing because investors poured tons of capital into them to subsidize the service.
Uber could not do taxis cheaper than existing if they didn’t have years of free cash to artificially lower prices.
We are in the beginning of late state capitalism, profitable companies go under due to private capital firms and absolute ponzi frauds get their faces on time magazine.
Enjoy the collapse.
I don’t know if you noticed this but some really big companies with high stock valuations are only existing because investors poured tons of capital into them to subsidize the service.
Exactly, they PAID MONEY to make it work. No they don’t make the money back and depend on outside capital, but they are still paying their employees (not enough) and suppliers, etc.
Yes, we are in late stage capitalism where the market eats itself.
Why do you think we have seen so much large scale fraud in the last 15 years?
Big Company: Well if you can’t afford food you should not have food.
Also Big Company:… sobbing pwease we neeed fweee… pwease we need mowe moneys!
You don’t get to both ignore intellectual property rights of others, and enforce them for yourself. Fuck these guys.
Serious Question: When an artist learns to draw by looking at the drawings of the masters, and practicing the techniques they pioneered, are the art students respecting the intellectual property rights of those masters?
Are not all of that student’s work derivative of an education based on other people’s work who will never see compensation for that student’s use?
I agree with you on principle. However… How long do you think it will be until these very same “AI” companies copyright and patent every piece of content their algorithms spew out? Will they abide by the same carve-outs they want for themselves right now? Somehow I doubt it.
They want to ignore the laws for themselves, but enforce them onto everyone else. This “Rules for thee but not for me” bullshit can’t be allowed to pass. Let’s then abolish all copyright, and we’ll see how long these companies last when everyone can just grab their stuff “for learning”.
How long before a self-owned AI company that does every administrative job better than humans because it trained on human behavior for 100 years?
What do you think an entity like that would be capable of?
A bit off-topic, but I’d be fine with that. The more mind-numbingly dumb work that computers can do for us, the less time we have to spend doing it ourselves. Administrative jobs holders disagree with this, but so did every person whose job and livelihood was replaced by automation, ever. UBI (universal basic income) is the only answer that will save all of us from starvation when automation eventually replaces us too.
I agree with everything in your post but the simple truth is administrative jobs are the modern equivalent of fluff court positions handed out to the 2nd+ born children of nobles and the modern owner class will never give up that eternal source of easy wealth.
Which is also why they fight so hard to keep anyone not in the owner class out of management.
One, let’s accept that there is a public domain, and cribbing freely from the public domain is A-OK. I can reproduce Michaelangelo all I want, and it’s all good. AI can crib from that all it wants.
AI can’t invent. People can invent: i can have a wholly new idea that no one has ever had. AI does nothing but recombine other existing ideas. It must have seed data, and it won’t create anything for which it has no initial input: feed it photographs only, and it can’t create a pencil drawing image. Feed it only black and white images, and it can’t create color images.
People do not require cribbing from sources. Give a toddler supplies, and they will create. So, we have established that there is a fundamental difference between the creation process. One is dependent on previous work, and one is not.
Now, with influences, you can ask, is your new creation dependent on the previous creation directly? If it is so utterly dependent on the prior work, such that your work could not possibly exist without that specific prior art, you might get sued. It will get debated and society’s best approximation of a collective rational mind will determine if you copied or if you created something new that was merely inspired by prior art.
AI can only create by the direct existence of prior art. It fakes invention. Its work has to come from somewhere else.
People have shown how dependent it is on its sources with prompts that say things like, “portrait of a patriotic soldier superhero” and it comes back with a goddamned portrait of Chris Evans. The prompt did not include his name, or Captain or America, and it comes back with an MCU movie poster. AI does not create. People create.
I think there is a fundamental difference here. People are not corporations. People have always learned like this and will always learn like this. Do we really want to allow large corporations to take knowledge from people, then commercialize it and put these very same people out of work?
Your distinction is mostly philosophical. Legally corporations have more protections than people.
I’m probably one of the most anti-corporate people you’ll meet today, I don’t even think publicly traded companies should exist.
To me, this reads like “Giant-ATV-Based Taxi Service Couldn’t Exist If Operators were Required to Pay Homeowners for Driving over their Houses.”
If a business can’t exist without externalizing its costs, that business should either a. not exist, or b. be forced to internalize those costs through licensing or fees. See also, major polluters.
“Ai” as it is being marketed is less about new technical developments being utilized and more about a fait accompli.
They want mass adoption of the automated plagiarism machine learning programs by users and companies, hoping that by the time the people being plagiarized notice, it’s too late to rip it all out.
That and otherwise devalue and anonymize work done by people to reduce the bargaining power of workers.
They also don’t care if the open, free internet devolves into an illiterate AI generated mess, because they need an illiterate populace that isn’t educated enough to question it anyway. They’ll still have access to quality sources of information, while ensuring the lowest common denominator will literally have garbage information being fed to them. I mean, that was already true in the sense that the clickbait news outsold serious investigative news, and so the garbage clickbait became the norm and serious journalism is hard come by and costly.
They love increasing barriers between them and the rest of the populace, physically and mentally.
Silicon valley’s core business model has for years been to break the law so blatantly and openly while throwing money at the problem to scale that by the time law enforcement caches up to you your an “indispensable” part of the modern world. See Uber, whose own publicly published business model was for years to burn money scaling and ignoring employment law until it could drive all competitors out of business and become an illegal monopoly, thus allowing it to raise prices to the point it’s profitable.
Fucking scooters lying all over the sidewalk.
Didn’t read the article but boo-fucking-hoo. Pay the content creators.
Free for me, paid by thee
This is not actually true at all, you could train very good LLMs on public domain only info, especially science oriented ones.
But what people want is a chatbot that can call on current events, and that is where the cost comes in.
Yup. Same as the way the rest of use and learn from the internet. We basically wouldn’t have the internet as we know it if it weren’t 99% free content.
“today’s general-purpose AI tools simply could not exist” … “as a profitable venture”
Well how about consent at the very least?
Data Leak at Anthropic Due to Contractor Error
TL;DR - Anthropic had a data leak due to a contractor’s mistake, but says no sensitive info was exposed. It wasn’t a system breach, and there’s no sign of malicious intent.
Most things that I could talk about were already addressed by other users (specially @OttoVonNoob@lemmy.ca), so I’ll address a specific point - better models would skip this issue altogether.
The current models are extremely inefficient on their usage of training data. LLMs are a good example; Claude v2.1 was allegedly trained on hundreds of billions of words. In the meantime, it’s claimed that a 4yo child hears something between 45 millions and 13 millions words through their still short life. It’s four orders of magnitude of difference, so even if someone claims that those bots are as smart as a 4yo*, they’re still chewing through the training data without using it efficiently.
Once this is solved, the corpus size will get way, way smaller. Then it would be rather feasible to train those models without offending the precious desire for greed of the American media mafia, in a way that still fulfils the entitlement of the GAFAM mafia.
*I seriously doubt that, but I can’t be arsed to argue this here - it’s a drop in a bucket.
The thing is, i’m not sure at all that it’s even physically possible for an LLM be trained like a four year old, they learn in fundamentally different ways. Even very young children quickly learn by associating words with concepts and objects, not by forming a statistical model of how often x mingingless string of characters comes after every other meaningless string of charecters.
Similarly when it comes to image classifiers, a child can often associate a word to concept or object after a single example, and not need to be shown hundreds of thousands of examples until they can create a wide variety of pixel value mappings based on statistical association.
Moreover, a very large amount of the “progress” we’ve seen in the last few years has only come by simplifying the transformers and useing ever larger datasets. For instance, GPT 4 is a big improvement on 3, but about the only major difference between the two models is that they threw near the entire text internet at 4 as compared to three’s smaller dataset.
My point is that the current approach - statistical association - is so crude that it’ll probably get ditched in the near future anyway, with or without licencing matters. And that those better models (that won’t be LLMs or diffusion-based) will probably skip this issue altogether.
The comparison with 4yos is there mostly to highlight how crude it is. I don’t think either that it’s viable to “train” models in the same way as we’d train a human being.
