It's because it doesn't matter how copyrighted material ends up in a project. If an LLM reproduces copyrighted material (which is very hard to verify) then the lawsuit from the copyright owner can still sink even robustly funded projects.
The fact is, open source has much more liability than closed source software does. If copyrighted material ends up inside a private code base it'll be nearly impossible for the owner to discover that and sue.
> If an LLM reproduces copyrighted material (which is very hard to verify) then the lawsuit from the copyright owner can still sink even robustly funded projects.
Do you have many examples of this actually happening that you could share?
I really don't see how this issue is going to be feasible for courtrooms to deal with in a world where big tech are bragging about large percentages of all their code being produced by LLMs.
In the 90s aircraft manufacturers basically stopped whole segments of the market (anything smaller than a piston twin) due to litigation. I wouldn't be horribly surprised to find we spend the next 10 years fighting about this stuff in court.
I'm fully aware of lots of times powerful companies have used the courts to protect their business. My point is basically all of the powerful companies are elbow deep in LLMs to the extent of some sort of mutually assured destruction if they were to start court cases like this.
LLMs have been firing out a lot of code for a year or 2 now at least, so I would have thought we'd start seeing some court cases by now?
It's purely conjecture at this point but my gut feeling is most companies are in high spirits about the costs they can save and the benefits it can bring to their own business. I don't think enough time has passed to see really high profile losers. What happens when Infosys starts losing money or Free CAD becomes the next SolidWorks due to LLM driven improvements?
The lack of modern small airplane designs, especially engines, is the extensive and thus expensive certification process. Most of what is a "new" airplane usually is derived from a very old, grandfathered design.
The only real innovation for small planes is in the electric flight area - when you gotta do a full certification anyway as you're piloting an entirely new drivetrain, you can financially afford to be much more innovative on other parts of the design.
Why wouldn't it be feasible for Apple, with an unlimited war chest, to go after Asahi?
I am not saying they will, but it is certainly possible for them.
Because if it's feasible for any company with a war chest to start a court case about a competitor maybe having a matching line of code to theirs via an LLM, then basically every single company with a warchest would be at war with eachother, because they're all using LLMs.
Business and code production would grind to a halt whilst basically every big tech company shares it's entire codebase with every other tech company for discovery. It's basically MAD.
And if it was feasible, given we've had a couple of years of all the big tech companies heavily using LLMs, there should be some notable court cases by now, surely?
> Do you have many examples of this actually happening that you could share?
If by "this" you mean "an LLM [reproducing] copyrighted material", I have one here [0], with the challenge posed and plagiarized response at [1], found via [2].
One wonders how often the "Don't plagiarize, make no mistakes about this!" instruction fails and the LLMs include nontrivial chunks of other people's work into what they emit.
> If by "this" you mean "an LLM [reproducing] copyrighted material
No. I have absolutely no doubt that happens, and I'm not defending it or encouraging it.
My point is whilst it technically might be illegal and happening all the time, if it's unenforceable or sets a precedence that would severely break the business world, there's every chance nobody would dare bring it to a court room.
I'm not aware of it happening yet (someone trying to enforce copyright on code that was put in production via an LLM) and I would have thought if it did, we would all know about the precedence now.
> ...someone trying to enforce copyright on code that was put in production via an LLM...
AIUI, in the US something that's entirely machine-generated is not eligible for copyright protection. If one could demonstrate that that machine-generated output is plagiarized human work and were rich enough to bring it to court, and able to wait five to ten years for the outcome, I have to believe that the usual copyvio rules would apply because that's not machine-generated output, it's straight-up unauthorized copying performed by a machine.
> My point is whilst it technically might be illegal and happening all the time...
If it's illegal, it's illegal. Refusal to enforce the law doesn't make the action any less illegal. Criminals who get away with their crimes are still criminals. [0]
> ... if it's unenforceable or sets a precedence that would severely break the business world, there's every chance nobody would dare bring it to a court room.
Or the highest court of the land would find a way to misinterpret "related" historical decisions to make the crime retroactively legal, yeah.
> ...I would have thought if it did, we would all know about the precedence now.
I'm not so sure. For one thing, the courts move really slowly when they're not very motivated to address something. For another, news that paints the twin VC darlings and their "industry" as villainous has a tendency to get buried by any one of a billion hype pieces or minor scandals that they have waiting in the wings.
[0] To bystanders who might wish to retort: Yes, I'm very aware that some things that are illegal should not be. I'm also aware that some things that are not illegal very much should be.
Can you state what actual point you're trying to make, rather than just picking each of my sentences and saying you're not so sure or you disagree? I could do the same to you, but it's just a waste of time if we're not trying to actually come to a conclusion together.
My hypothesis is that I suspect there's a MAD type situation where technically - by the letter of the law - all the big tech companies bragging about x% of their code being LLM generated are basically admitting to breaching copyright laws. But if one well funded tech company successfully prosecutes that and sets a precedence, then all big tech is going to have to wind back all the LLM code its put in production in the past few years to prevent litigation, which is probably unworkable, hence MAD.
I was looking to disprove my own hypothesis by asking for court cases prosecuting this. I'm not interested in arguing semantics with you.
(FWIW, I was repeatedly taught by various legal scholars at various levels of education that a criminal is someone who has been found by a court to be guilty of committing an act that violates criminal law. Up until that point, they're usually just a suspect or similar. Not debating, just highlighting that your definition of criminal doesn't negate or disprove any point I'm making. Same for your definition of illegal.)
If you're looking for a single point, that's going cause you trouble. Maybe go back and read what I wrote more carefully? I believe that I took care to make sure that my rebuttals to your claims both dovetailed in nicely with the part of the claim they were rebutting and -as a backup- contained the context needed to understand exactly what I was rebutting.
> I was looking to disprove my own hypothesis by asking for court cases prosecuting this.
_Prosecuting_? If you know enough to ask for that, then you have the knowledge needed to discover that those court cases don't really exist yet. The courts move slowly, especially when wealthy and/or powerful entities are supremely disinterested in attracting their attention.
> I'm not interested in arguing semantics with you.
For anything that's not incredibly clear-cut, that's like the entirety of the practice of law... at least in the US. At its heart, it's an adversarial storytelling exercise that has hundreds to thousands of years of tradition to pull from. I know this in part because I strongly considered becoming a lawyer before I became a programmer.
> ...a criminal is someone who has been found by a court to be guilty...
That's very good practice to engage in when talking about ordinary people. When talking about very powerful people or entities (such as large, influential companies), it's not so great. Very powerful entities tend to be permitted to get away with antisocial conduct that the rest of us get metaphorically nailed to the wall for. "Rules for thee but not for me" is trite, but very, very frequently how things end up working.
EDIT: There's also the example that I had in mind yesterday of US-based marijuana growers, commercial purchasers, individual purchasers, and consumers. All of these people are knowingly illegally producing, trafficking, and/or consuming a Schedule I substance. Folks often do this in the open, making no secret of it. The DoJ and DEA might choose not to haul these people into court, but most -if not all- of them are still very plainly criminals... just not convicted ones.
> My hypothesis is that I suspect there's a MAD type situation...
No. In regards to this whole LLM craze, there's no MAD-type situation between well-funded tech companies. There are a handful of players who have absolutely no interest in stopping the money-making party. No one involved in the money-making party that resulted in the 2008 crash who could make a credible report to the relevant authorities had any interest in calling those authorities in to stop it... why on earth would they when there's so much money being made? And since approximately none of the folks responsible for that huge pile of fraud saw any meaningful punishment, why wouldn't businesses with the means to engage in activity that's no less illegal but is very profitable for a large number of powerful people be discouraged from doing it again?
The fact is, open source has much more liability than closed source software does. If copyrighted material ends up inside a private code base it'll be nearly impossible for the owner to discover that and sue.