Thomson Reuters Wins First Major AI Copyright Case In the US 54
An anonymous reader quotes a report from Wired: Thomson Reuters has won the first major AI copyright case in the United States. In 2020, the media and technology conglomerate filed an unprecedentedAI copyright lawsuit against the legal AI startup Ross Intelligence. In the complaint, Thomson Reuters claimed the AI firm reproduced materials from its legal research firm Westlaw. Today, a judge ruled (PDF) in Thomson Reuters' favor, finding that the company's copyright was indeed infringed by Ross Intelligence's actions. "None of Ross's possible defenses holds water. I reject them all," wrote US District Court of Delaware judge Stephanos Bibas, in a summary judgement. [...] Notably, Judge Bibas ruled in Thomson Reuters' favor on the question of fair use.
The fair use doctrine is a key component of how AI companies are seeking to defend themselves against claims that they used copyrighted materials illegally. The idea underpinning fair use is that sometimes it's legally permissible to use copyrighted works without permission -- for example, to create parody works, or in noncommercial research or news production. When determining whether fair use applies, courts use a four-factor test, looking at the reason behind the work, the nature of the work (whether it's poetry, nonfiction, private letters, et cetera), the amount of copyrighted work used, and how the use impacts the market value of the original. Thomson Reuters prevailed on two of the four factors, but Bibas described the fourth as the most important, and ruled that Ross "meant to compete with Westlaw by developing a market substitute." "If this decision is followed elsewhere, it's really bad for the generative AI companies," says James Grimmelmann, Cornell University professor of digital and internet law.
Chris Mammen, a partner at Womble Bond Dickinson who focuses on intellectual property law, adds: "It puts a finger on the scale towards holding that fair use doesn't apply."
The fair use doctrine is a key component of how AI companies are seeking to defend themselves against claims that they used copyrighted materials illegally. The idea underpinning fair use is that sometimes it's legally permissible to use copyrighted works without permission -- for example, to create parody works, or in noncommercial research or news production. When determining whether fair use applies, courts use a four-factor test, looking at the reason behind the work, the nature of the work (whether it's poetry, nonfiction, private letters, et cetera), the amount of copyrighted work used, and how the use impacts the market value of the original. Thomson Reuters prevailed on two of the four factors, but Bibas described the fourth as the most important, and ruled that Ross "meant to compete with Westlaw by developing a market substitute." "If this decision is followed elsewhere, it's really bad for the generative AI companies," says James Grimmelmann, Cornell University professor of digital and internet law.
Chris Mammen, a partner at Womble Bond Dickinson who focuses on intellectual property law, adds: "It puts a finger on the scale towards holding that fair use doesn't apply."
Chris Mammen, a partner at Womble Bond Dickinson (Score:2)
Womble? Come on...
Re: Chris Mammen, a partner at Womble Bond Dickins (Score:2)
Proof that this is a simulation and our world was generated with AI ;)
Re:Disaster for the little guy (Score:5, Insightful)
Re: (Score:2)
Exactly. No idea why people confuse this.
Re:Disaster for the little guy (Score:4, Insightful)
They confuse them because they can. Brute force, breaking things, these are all how the tech bros muscle their way in, disrupt, steal, and become immune because of their ability to put the legal system on constant offense. It's a well-known strategy, used because it works.
Kleptocracy is more profitable. Dodging taxes because you bribed the tax law goes along with it, the multi-jurisdictional game. Exporting labor costs, and/or amusingly shifting to your shaky AI is still another strategy (thanks, Meta!, who'll be the next one to face this judgment).
Re: (Score:2)
I think one of the few really good outcomes of this AI stuff is it might break the copyright system.
But more likely they'll carve out a place for huge corporations to use different standards for fair use than the individual and we'll just be even more screwed than before.
Re: (Score:3)
Or, they could pay the authors of the book real compensation for their hard work; often, their publishers cave in and just license an entire catalog, authors, like musicians, be damned.
They get screwed on the actual work side, authors, poets, musicians, artists, while the profit rises to the top to be embedded in some hallucinating AI.
Re: (Score:1)
The Big Tech AI providers will all get sued by copyright holders, will enter into bulk licensing agreements, and will then be fine to continue. Anyone running their own smaller models or trying to do any of the same stuff outside of "Big AI" will not be able to afford the endless lawsuits or legitimate licensing agreements and will thus be forced to stop. The end result will be that Big AI will be able to use this as a competitive moat to keep any smaller lesser-funded competitors out of the business altoge
Re: Disaster for the little guy (Score:4, Informative)
It was also invented for the purposes of critique and journalism, even if those things are done by corporations.
That still doesn't apply to this necessarily, but it is a LOT wider than you portray it.
Re: Disaster for the little guy (Score:4, Informative)
The act of stuffing information into an AI model is not in itself directly critique or journalism so that cannot apply here. If all they were doing was processing information through an algorithm that directly created articles as its output they may have an argument for this.
What they are actually doing is using someone else's database that they haven't licensed to build a new commercial database with an unconventional interface and data storage that may or may not then be used to write text that may or may not be considered journalism.
Re: (Score:3)
But... they're not going to remain billionaires unless they're allowed to break copyright here!
Re: Disaster for the little guy (Score:1)
Fair use is for everyone.
This really isnâ(TM)t that hard. Fair use is not â" nor was it ever intended to be â" a backdoor âoepay what you can affordâ.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:3)
Bibas described the fourth as the most important, and ruled that Ross "meant to compete with Westlaw by developing a market substitute."
Pretty much every company that ingests copyrighted material is looking to eventually reproduce it by developing a market substitute. Did you write a love poem? Guess what: the LLM, when prompted to write a love poem, will draw on yours. Did you write a scientific paper about DNA? Guess what: the LLM, when prompted to write a paper about DNA, will be influenced by yours.
Therefore, *every* LLM that uses copyrighted material is in violation, since it can be used to compete with the original, copyrighted ide
Boon for Chinese AI companies? (Score:4, Interesting)
If the lawsuits block US AI companies, what is to prevent Chinese AI companies from completely ignoring them?
I can see the US AI companies facing serious challenges in the not-so-distant future. The Chinese already have a labor and cost advantage. I am reminded of what the Japanese did to the US electronics industry.
Re: (Score:2, Troll)
If we're now allowed to punch neighbors we disagree with, then what prevents the mafia from doing this? In other words, the attitude that one should be allowed to break the law because other people are doing it doesn't hold up.
Re: (Score:2)
International copyright law. The thing that Chinese companies rely heavily on to protect their intellectual property.
The Japanese didn't steal anything from the US electronics industry. They started out making basic components, developed their own ICs, and licenced US designs. Famously NEC licenced Intel's 8086, and then improved it and got sued. They won in US court - their licence was deemed to give them the right to produce 286 and 386 compatible parts as well. They also did the R&D to get ahead, e.g
Sounds right (Score:4, Interesting)
Either AI companies are in bullshit dreamland, or copyright falls.
Re: (Score:2)
This is the beginning of the end for chatbots that simply tell you what you want to know, instead of constantly skewing the results or trying to distract you in order to turn a buck.
Not about generative AI, per se (Score:5, Informative)
This involved human beings taking summaries from a database, rewriting them negligibly, and then loading that modified content into generative AI, and doing so in explicit violation of rules saying not to use the database for that purpose, with the explicit intent to build a service that competes against the service provided by the original database. This isn't a particularly broad decision.
This is basically the phone book copying case, but without the factuality defense. Had they started with the original case notes instead of the other company's summaries, they probably would have been in the clear, but they did not.
Re: (Score:1)
Re: (Score:1)
If you're worried about infringing the "first summarizer" t
Fair use should only apply... (Score:3)
To companies that release fully uncensored and unrestricted open models and weights for free use and redistribution.
what was the copyright violation? (Score:3)
The article didn't even say what the alleged copyright violation was. The fair use doctrine doesn't matter until we know what the use actually is.
Artificial neural networks are trained by "reading" text, just like our brains are trained by reading text. If the claim is that training an AI by reading text is not "fair use" because training the AI is an intention to create a competitor, then that text cannot be read by humans either. Now, it could be argued that humans do not intend to "create a competitor" by reading text, but law students certainly do. Can law students, or attorneys, NOT read legal documents because it is not "fair use"? The usage here is very important, if the argument is simply that training is not fair use then there is going to be a big problem.
Re: (Score:1)
I looked at the actual decision, it seems really crappy to me. It doesn't appear to me to say what the "use" actually was. It would seem that the actual "copyright violation" was made by a third party prior to involvement by any AI at all. No consideration of the AI itself is even made. Instead, the judge only seems to care that copyrighted materials were fed into a system intended to be a commercial product. Sure, but attorneys practicing law do that too. My comment remains, can law students read cop
Re: what was the copyright violation? (Score:3)
No, he definitely does not. Software is not a person.
Re: (Score:2)
No, he definitely does not. Software is not a person.
Don't give the supreme court ideas!
Re: (Score:2)
Software is not a person and the weights that result from training this software are nothing but a lossy encoding of the training data.
The judge is right to only care that they fed copyrighted material into the software and disregard any attempt at drawing parallels to attorneys practicing law.
Re: (Score:2)
My comment remains, can law students read copyrighted legal writing pursuant to their law degrees? This judge says no.
The law recognizes people as a special case, even if you do not. LLMs are machines, not people. Your "argument" is essentially a hallucination and has no merit.
Re: (Score:1)
"In other words, Ross built its competing product using Bulk Memos, which in turn were built from Westlaw headnotes."
It can't be a competing product unless the human lawyers can get summaries out of it. Which the judge then found, when comparing to the original summaries to which access was established, substantial similarity sufficient to decide infringement
Re:what was the copyright violation? (Score:4, Insightful)
So read a lot of copyright works, memorize it exactly, then you regurgitate it almost exactly and try to sell that. This is not fair use, even for a human. At the very least, summarize without copying. And yes, in legal arenas this means summarize the summaries if the summaries are copyrighted.
Re: (Score:2)
Re: (Score:2)
Yeah but laws happen to be public documents that are meant to be copied and learned, so no copyright violation there.
Also it becomes a life profession to have humans dedicate that level of effort to achieve such memorization.
Re: (Score:2)
Interpretations of the law, written down into books, are copyrighted though. As opposed to actual legal rulings. That's the snag in this particular story.
Re: what was the copyright violation? (Score:3)
Adding the automation fundamentally changes the scale of how far and wide and fast an idea can go.
I think its pretty obvious that harvesting information on an industrial scale with intent to monetize is quite obviously not fair use.
Re: (Score:2)
"I think its pretty obvious that harvesting information on an industrial scale with intent to monetize is quite obviously not fair use."
I agree UNLESS you are making it copyleft or bsd type open and widely distributing. It really doesn't matter if you attempt to monetize if you've rendered the advantage into the commons.
Re: (Score:2)
The problem there is that, for several years now, there has been a narrative which has even made inroads into Slashdot that "but thEy Did IT On A COmpUteR" means nothing and that the law should be applied as though the computer does not fundamentally change the action and should be treated as if it were a person under the law. There have been legal cases already that were adjudicated with that very conclusion. You seem to want to go with the alternative conclusion that using a computer to do something DOE
Re: (Score:2)
We don't have to treat artificial computing systems deliberately designed to encode and store training data as weights in the same way as real brains and therefore parallels between their operation don't matter. Copyright is artificial, brains are not and predate copyright... a double standard on that basis is just fine.
Fair Use is intended for individual's private use. (Score:3, Interesting)
Perhaps libraries and such also have fair use protections, but that is about it.
Re: (Score:2)
Fair use very often about something bigger than the individual. Fair use does cover parodies, even if those parodies are for other people and those parodies make money. Also being allowed to copy a couple of sentences into a news article or your term paper.
The GOP will save the AI firms (Score:3, Insightful)
Elon Musk is the biggest donor to the Republican Party. He also owns an AI company. Heâ(TM)s best buddies with the President and the GOP congresscritters defer to him. There will be a law passed allowing AI companies to do whatever they want with copyrighted works. Theyâ(TM)ll tell us that itâ(TM)s to keep China from gaining AI supremacy blah blah national security blah blah and people will lap it up.
Re: (Score:3)
I fear this is all very accurate. Why would trump care about fair use?
Funny thing is elon's Ai will probably be a dud and get stomped by the better companies.
Re: (Score:2)
And, let me think, you expect this "law" to apply globally? Why would it? That is if it is found to be constitutional in the first place. And they are up against Disney here.
This ruling essentially predicts what is going to hold all over the place, namely that training data was obtained in a massive criminal piracy campaign by almost all model creators.
Quite right too .. (Score:4, Interesting)
Skittles (Score:2)
die pricks (Score:1)
it's obviously been time to eliminate the entire intellectual property and patent system, or at least reduce its use case to ART AS WAS INTENDED, and stop letting these greedy fuckheads run ripshod over us for decades now. Them getting rich is not worth us getting fucked out of the ability to do things right.