Sarah Silverman’s AI Case Isn’t Going Very Well Either
from the stop-trying-to-make-copyright-stop-ai dept
Just a few weeks ago Judge William Orrick massively trimmed back the first big lawsuit that was filed against generative AI companies for training their works on copyright-covered materials. Most of the case was dismissed, and what bits remained may not last much longer. And now, it appears that Judge Vince Chhabria (who has been very good on past copyright cases) seems poised to do the same.
This is the high profile case brought by Sarah Silverman and some other authors, because some of the training materials used by OpenAI and Meta included their works. As we noted at the time, that doesn’t make it copyright infringing, and it appears the judge recognizes the large hill Silverman and the other authors have to climb here:
U.S. District Judge Vince Chhabria said at a hearing that he would grant Meta’s motion to dismiss the authors’ allegations that text generated by Llama infringes their copyrights. Chhabria also indicated that he would give the authors permission to amend most of their claims.
Meta has not yet challenged the authors’ central claim in the case that it violated their rights by using their books as part of the data used to train Llama.
“I understand your core theory,” Chhabria told attorneys for the authors. “Your remaining theories of liability I don’t understand even a little bit.”
Chhabria (who you may recall from the time he quashed the ridiculous copyright subpoena that tried to abuse copyright law to expose whoever exposed a billionaire’s mistress) seems rightly skeptical that just because ChatGPT can give you a summary of Silverman’s book that it’s somehow infringing:
“When I make a query of Llama, I’m not asking for a copy of Sarah Silverman’s book – I’m not even asking for an excerpt,” Chhabria said.
The authors also argued that Llama itself is an infringing work. Chhabria said the theory “would have to mean that if you put the Llama language model next to Sarah Silverman’s book, you would say they’re similar.”
“That makes my head explode when I try to understand that,” Chhabria said.
It’s good to see careful judges like Chhabria and Orrick getting into the details here. Of course, with so many of these lawsuits being filed, I’m still worried that some judge is going to make a mess of things, but we’ll see what happens.
Filed Under: copyright, joseph saveri, llama, matthew butterick, sarah silverman, vince chhabria
Companies: meta, openai
Comments on “Sarah Silverman’s AI Case Isn’t Going Very Well Either”
Might Generative AI Infringe Copyright?
There are some interesting and more nuanced arguments worth perusing on this subject. TI recommend this Harvard Law School webinar https://www.youtube.com/watch?v=0p-mzF2IBiM in which Pamela Samuelson (who, I assume, most everyone is familiar) debates with Justin Hughes, an i.p. law professor who I hadn’t come across previously. Pamela argues for the position that generative AI generally does not infringe but Justin has some interesting counterarguments.
Re:
If it infringes, that’s an indictment of copyright law.
Re:
Justin Hughes being a well known copyright maximalist, who has never seen a copyright law he didn’t want to extend further, and who regularly argues against fair use.
Oh, thank god.
If AI had a bibliography, said bibliography page count might exceed that of the work itself.
Re:
Not “might”. “Would”.
“I’m still worried that some judge is going to make a mess of things, but we’ll see what happens.”
The problem is that one gullible judge will become the one out of a thousand case that becomes precedent and there’d be no way that he could undo his previous ruling. Sure do wish there was a way that judges could know that a plaintiff is trying to staff-split and get an affirmative answer after being denied umpteen times. That way, the judge’s time is saved and he doesn’t have to become the gateway by which a million formerly innocent people have to deal with these trials.
If I read the book and gain knowledge from the book am I infringing on its copyright?
The fact that an AI may have more unchangeable memory than I do shouldn’t matter if it’s reading the work or if I’m reading the work.
Re:
However the logic behind these cases is someone has read my book and is now making money, so make them give me a lot of money.
Re: Re:
I had that thought before you posted it! I deserve compensation!
Re: RE: If I read the book and gain knowledge from the book am I infringing on its copyright?
This was my thought too. What is an AI doing with previous works that a human does not do, other than remember really well? After all, every book is inspired by the conglomeration of previous literature and experiences the author has had. The only difference is that Ms. Silverman’s contribution is far more miniscule since the AI has had the opportunity to consume far more and more varied works to feed its “creativity”.
Re: Re:
Now there’s an idea equal parts absurd and disturbing, using the same arguments against AI one could argue that if a human being with sufficiently good memory read a book then possibly their existence(since they claimed that the AI itself is infringing) and certainly any works they write would automatically be treated as infringing on copyright.
love this line
The authors also argued that Llama itself is an infringing work. Chhabria said the theory “would have to mean that if you put the Llama language model next to Sarah Silverman’s book, you would say they’re similar.”
Re: Potential Sources of Liability
Without commenting specifically on the Silverman case, I think it remains worth a look at the higher level issue. Obviously a Llama system is not substantially similar to the i.p. of the plaintiff. With very high probability, the outputs of the Llama system are neither substantially similar to, nor derivative works of, the plaintiff’s i.p. (but beware corner cases). However, we generally do not know what went into building and training that system and it is possible that infringement occurs at some such earlier point that contaminates whatever happens afterward.
Re: Re:
What infringement could happen in the earlier part? And I mean an infringement that isn’t pure speculation.
Re: Re: depends on what you meant.
If you mean using copyrighted material to train with is infringing then no. If you meant they used pirated copies then maybe, but no one has shown any evidence that ever happened.
Re: Re: Re:
AFAIK, no one has ever been accused of copyright infringement for reading a “pirated” book.
The basic premise of these cases is essentially “but on a computer”, see patents were someone takes a mundane task a human does and applies for a patent when done by a computer.
Re: Re:
The only thing being infringed, at best, is contract law, and even then, that’s a very vague thing and only happens at the collation of training materials level.
Anything else that happens after is not and will never be copyright infringement, and to claim otherwise is to be anti-education.
…Also, you do realize that by claiming that READING AND WRITING is copyright infringement, you basically are implying that people can’t read or write without “prior permission”, yes? How do you justify people being unable to do just about anything worthwhile? Like, I dunno, actually talk to one another? Or do business? Or READ THE DAMN BILLS THEY ARE SUPPOSED TO PAY?
Man, the Bayside Advisory simp from the June 2022 article was a riot to read. “If you don’t agree with copyright holders it means you’re a closet minimalist and pirate!” Classic, average_joe-level, trash-tier pro-IP logic.
Getting a little desperate there...
The authors also argued that Llama itself is an infringing work. Chhabria said the theory “would have to mean that if you put the Llama language model next to Sarah Silverman’s book, you would say they’re similar.”
By the argument of ‘knowing about and/or learning from a work means you’re infringing upon it’ copyright infringement becomes almost as common as air and stupidity as every person who has ever read a book and remembers any of it is engaging in copyright infringement.
If this is really the best that authors and publishers can come up with to argue against AI then they have less than nothing and are just making themselves look like self-entitled fools.
Re:
So, never read Sarah Silverman, got it.
AS well as the authors and publishers who are trying to sue these unethical random content generator companies.
Re: Re:
Yeah, if those guys can just list all the works they have so they can be avoided, that’d be great. Wouldn’t want to get randomly accused of infringement and all.
What about the inputs?
I agree that no one has put forth evidence of any specific wrongdoing in the creation of the models that might give rise to liability. Hughes (who I had no previously known about and who Mike tells us is a maximalist) does not claim the contrary–he just argues that it is worth looking into. I imagine that, in some future case, it will get looked into.
Re:
Sounds exactly like the arguments put forward by copyright lawyers. “We’re not saying that this child or grandparent downloaded our songs, we’re just demanding to look into all their subscriber information, ordered by the judge and sponsored by the judiciary, just in case there’s something we can use.”
Copyright and its enforcement seem to be entirely premised on fishing expeditions. It’s no wonder judges got sick and tired of it after Prenda Law tried it.
Hughes
Life is too short (mine anyway) to summarize the next level of detail of Hughes’ positions. Also, I’m on the fence about the optimal “right answer” to this set of issues. I encourage anyone who wants to dig deeper to watch the Youtube whose link I posted earlier. You’ll also get to hear Pamela Samuelson disagree with Hughes oin many points.
Re:
But someone else’s life isn’t too short, then?
You also seem to have very little issue with disputing what you think Hughes’ position is not.