GitHub Copilot copyright case narrowed but not neutered

The judge overseeing the AI code-copying case filed against GitHub, OpenAI, and Microsoft has dismissed some but not all of the aggrieved developers' claims, leaving the plaintiffs a more limited but still potentially potent opportunity to challenge the alleged algorithmic reproduction of their source code. In November 2022, …

[Anonymous Coward]

CloseMindedAI

The irony is if I asked a question about software for a particular problem, or described an algorithm in general terms hoping for some clues about implementations, I would value primarily usage documentation and references to usage of those implementations - at least 99% of the time. Almost always there are variations on implementations, and usage considerations to (re)load into the brain before committing to stitching in a particular piece of code (or adapting it).

Even a reference to a snippet from stack overflow would likely benefit from the discussion surrounding it. It would just be stupid to not include it.

The best AI service seen so far is "Perplexity", which smartly place priority on giving actual (not made up) references - never a made up reference. Gee whiz, so obvious, huh? Close<Minded>AI can't or won't do that probably because their policy is a reflection of their sad leaders values - in court fighting for the right to be worthless.

[Groo The Wanderer]

If Co-Pilot "adheres to applicable laws" then the laws need to change, but I don't think that's the case: All three defendants clearly violate developer's copyright with impunity.

[Joe W]

Yeah?

" it is a modification: for instance, a near-identical copy that contains only semantically insignificant variations of the original Licensed Materials"

So changing the variable names removes any and all copyright? Good to know... I was under the impression that me porting code e.g. from Matlab to R would be minor enough changes to retain the original license. And now they tell me I can just rename variables and switch to tab indentations and I'm in the clear?

[doublelayer]

Re: Yeah?

I'm looking forward to OpenAI's code leaking so I can perform these changes and make as much money as I want from it without paying them anything. Do you think they'll be happy when that happens?

[Falmari]

Re: Yeah?

@Joe W "So changing the variable names removes any and all copyright?"

That's a strawman argument. The judge was only dismissing the plaintiffs claims of DMCA violation, the "Removal or Alteration of Copyright Management Information." not the whole case.

The reason for the dismissal, the defendants argued that that DMCA violation can only be claimed for an identical copy of a copyrighted work. The judge agreed giving reasons and cases for why he agreed. Because the plaintiffs were claiming the output from Copilot that infringed their code was not identical, the examples they provided were not identical, they had insufficient grounds for that particular DMCA violation.

Your argument is a fallacy, the judge is not ruling that if you fail to meet the bar for DMCA violation, the "Removal or Alteration of Copyright Management Information" there can be no claims of copyright infringement, otherwise he would have dismissed the case and all claims for failing to meet said bar.

BTW to violate that DMCA requires it to be done intentionally or knowingly.

[Anonymous Coward]

Re: Yeah?

I guess Bots can't do anything intentionally or knowingly?

[Copyright bots and classical musicians are fighting online. The bots are winning., Washington Post, May 2020]

A few Sundays ago, Camerata Pacifica artistic director Adrian Spence, aided by his tech-savvy son Keiran, went live on Facebook to broadcast a previously recorded performance of Mozart’s Trio in E flat (K. 498), a.k.a. the “Kegelstatt” trio. At least they tried to.

The recorded performance was one of many that Spence had drawn from the Camerata’s extensive video archives. When the covid-19 crisis abruptly canceled its season, Spence launched a weekly series of rebroadcasts to fill the silence. These broadcasts, even with their modest virtual attendance of 100 or so viewers per stream, have been essential to keeping Spence’s Santa Barbara-based chamber organization engaged with its audience.

That is, until that recent Sunday, when his audience started to disappear, one by one, all the way down to none.

[doublelayer]

Re: Yeah?

The provision that talks about keeping the information is pretty important. That information, for example, is the way that licenses make it clear they apply to the file. Changing the file does not remove that requirement, since most if not all open source licenses explicitly state that the license must remain with the file. If changing the file even slightly means that they can get out of that provision, then why can't they invalidate other ones on the same basis? The clauses concerned state that removing that data is illegal when it is linked to a violation, which is very similar language to most other clauses in the law. If you can mentally add "unless the file is even slightly different" to the end of that sentence, why can't I add it to all similar sentences that use the same conditions?

Obviously, this should not be the case. I think this judgement is clearly misinterpreting both the intent and the literal meaning of the legislation as written. You do not invalidate copyright when you change a file slightly. You do not cancel the requirement of the license to retain the license information with that file. You do not invalidate the law that ensures that you do not do so. This should be overturned.

[Pascal Monett]

“We firmly believe"

Oh well, case dismissed then, right ?

So, if I firmly believe that I have the right to take a bag of money from a bank I have no account in, then I'm in the clear ?

Somehow, I have a nagging doubt that my personal belief system is not an excuse. But that's PR for you . . .

[catprog]

Re: “We firmly believe"

No. The parts that were dismissesed work like this.

It is a crime to rob the bank.

It is a separate crime to take money out of account 476 at the bank.

The prosecution was saying they commited both crimes by taking money out of account 477.

The defense says we cannot be guilty of the second crime as the account in question is 477 not 476.

They still have to answer for the first accusation.

[johnrobyclayton]

Picking up free crap and selling it to the highest bidder is the American Dream

It would not be hard to train or hire a few hundred coders and get them to implament various algorithms in various languages.

Ensure the code is liberally commented.

Hire some more experienced coders/develpers to to verify that the coding meets various standards (security, effeciency, indentation, style, and anything else that might be useful)

From this, create a corpus of code with known provenance, known quality, known bias from which to train an AI.

You could also sell it over and over again to any other developers of other AIs.

It is just easier to get free crap, (stolen or otherwise) without any quality control, package it up with some buzz words and try to sell it to any rubes they can get to pay for it.

[imanidiot]

Re: Picking up free crap and selling it to the highest bidder is the American Dream

It's hard to implement anything even remotely creative or effective if it is not done in the pursuit of a specific outcome/goal. How to implement a sorting algorithm (or what algorithm to choose) can be very dependent on the source data and the desired outcome for instance. So just having "monkeys on typewriters" bashing out code, even if it's perfectly valid code, is unlikely to get you the sort of goal oriented, clever, efficient, and novel implementations that code "in the wild" is going to give. Which is why companies like GitHub are so eager to wholesale ingest anything and everything from sites like StackOverflow where actually smart people come to discuss smarter and more efficient ways of solving their problems.

[Groo The Wanderer]

Re: Picking up free crap and selling it to the highest bidder is the American Dream

If this were feasible, there would have been businesses SELLING those "do it all" packages for very great profit before OpenAI ever sparked a neuron to fire in anyone's brain.

[Carlie J. Coats, Jr.]

Judge ignored the law: Derivative works

From the Copyright Act (17 USC section 101):

A “derivative work” is a work based upon one or more preexisting works, such as a translation, musical arrangement, dramatization, fictionalization, motion picture version, sound recording, art reproduction, abridgment, condensation, or any other form in which a work may be recast, transformed, or adapted. A work consisting of editorial revisions, annotations, elaborations, or other modifications which, as a whole, represent an original work of authorship, is a “derivative work”.

The "Judge Tigar in his order agreed with the defendant companies' argument that the alleged DMCA violations can only occur when the copied works are identical" is completely contrary to that section of the law.

The "The judge also tossed the plaintiffs' claims under Sections 1202(b)(1) and 1202(b)(3) of the US Digital Millennium Copyright Act" is also in complete defiance of the Law.

Judge Tigar should be impeached.