Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

We've filed a lawsuit against GitHub Copilot

githubcopilotlitigation.com

Post reply on HN
Print view View original

281–290 of 824 posts

Re: We've filed a lawsuit against GitHub Copilot

#281

Post by jacobr1 »

adlpz wrote: post #189
nrb wrote: post #160

Earlier quoted context omitted.

Does anyone have a problem with it, so long as the material it trained on was with explicit permission/license and not potentially in violation of copyright? That's where the line is for it to be suspect IMO.

I guess I'm just afraid that it might not be as good as it is that way. It's a bit like how GPT-3, Stable Diffusion and all those generative models use extensive amounts of copyrighted material in training to get as good as they do. In those cases however the output space is so vast that plagiarism is very unlikely. With code, not so much.

GPT-3 and Stable Diffusion might not copy things exactly - but they certainly do copy "style" There are many articles likes this:

https://hyperallergic.com/766241/hes-bigger-than-picasso-on-...

The interesting thing is that the names get explicitly attached to these styles. It isn't exactly a copyright issue, but I'm sure it will get litigated regardless.

Re: We've filed a lawsuit against GitHub Copilot

#282

Post by bun_at_work »

IshKebab wrote: post #253
jacooper wrote: post #215

Earlier quoted context omitted.

No thank you. I put a license to be followed, not to just be disregarded by an AI as "Learning material". No human perfectly reproduces their learning material no matter what, but Copilot does.

So are you ok with it being illegal for humans to learn from copyrighted books unless they have a license that explicitly allows learning? That does not sound like a pleasant consequence.

AI are not humans, no human can read _all_ the code on Github. They certainly can't read _all_ the code on Github at the scale that MS can, and are unlikely to be able to extract profits directly from that code, in violation of the licensing.

Re: We've filed a lawsuit against GitHub Copilot

#283

Post by foooobaba »

lbotos wrote: post #152
foooobaba wrote: post #81

If github or google indexes source code using a neural net to help you find it, given a query, is that also illegal? If you think of copilot as something that helps you find code you’re looking for, is it all that different, and if so, why? In this case, wouldn’t the users of copilot be the ones responsible for any copyrighted code they may have accessed using copilot?

The crux of the issue: Is the code that is being generated being used in a way that it's license allows? That's it. I'm confident that this problem would go away if copilot said: //below output code is MIT licensed (source: github/repo/blah) And yes, the "users" are responsible, but it's possible that copilot could be implicated in a case depending on how it's access is licensed. Stable diffusion has this same proble…

This makes sense, it produces chunks not the whole source where a search engine would also give you the license.

Re: We've filed a lawsuit against GitHub Copilot

#284

Post by bjourne »

adlpz wrote: post #189
nrb wrote: post #160

Earlier quoted context omitted.

Does anyone have a problem with it, so long as the material it trained on was with explicit permission/license and not potentially in violation of copyright? That's where the line is for it to be suspect IMO.

I guess I'm just afraid that it might not be as good as it is that way. It's a bit like how GPT-3, Stable Diffusion and all those generative models use extensive amounts of copyrighted material in training to get as good as they do. In those cases however the output space is so vast that plagiarism is very unlikely. With code, not so much.

I think the prompt "GPT-3, tell me what the lyrics for the song Stan by Eminem is" is very likely to output copyrighted material. The same copyrighted material is, of course, already republished without permission on google.com.

Re: We've filed a lawsuit against GitHub Copilot

#285

Post by buzzy_hacker »

m00x wrote: post #52
buzzy_hacker wrote: post #33

Copilot has always seemed like a blatant GPL violation to me.

Care to explain in legal terms why this stance is qualified?

You may convey a work based on the Program, or the modifications to produce it from the Program, in the form of source code under the terms of section 4, provided that you also meet all of these conditions:

a) The work must carry prominent notices stating that you modified it, and giving a relevant date. b) The work must carry prominent notices stating that it is released under this License and any conditions added under section 7. This requirement modifies the requirement in section 4 to “keep intact all notices”. c) You must license the entire work, as a whole, under this License to anyone who comes into possession of a copy. This License will therefore apply, along with any applicable section 7 additional terms, to the whole of the work, and all its parts, regardless of how they are packaged. This License gives no permission to license the work in any other way, but it does not invalidate such permission if you have separately received it.

——

I don’t see how one could argue that training on GPL code is not “based on” GPL code.

Re: We've filed a lawsuit against GitHub Copilot

#286

Post by thomastjeffery »

r3trohack3r wrote: post #158

I'm not confident in this stance - sharing it to have a conversation. Hopefully some folks can help me think through this! The value of copyleft licenses, for me, was that we were fighting back against the notion of copyright. That you couldn't sell me a product that I wasn't allowed to modify and share my modifications back with others. The right to modify and redistribute transitively though the software license ga…

It looks like you're missing the entire purpose of copyleft vs public domain.

The point is that copyleft source code cannot be used to improve proprietary software. That limitation is enforced with copyright.

Proprietary software is closed source. You can't train your NN on it, because you can't read it in the first place.

If someone takes your open source code and incorporates it into their proprietary software, then they are effectively using your work for their private gain. The entire purpose of copyleft is to compel that person to "pay it forward", by publishing their code as copyleft. This is why Stallman is a proponent of copyright law. Without copyright, there is no copyleft.

Re: We've filed a lawsuit against GitHub Copilot

#287

Post by r3trohack3r »

flatline wrote: post #176
r3trohack3r wrote: post #158

I'm not confident in this stance - sharing it to have a conversation. Hopefully some folks can help me think through this! The value of copyleft licenses, for me, was that we were fighting back against the notion of copyright. That you couldn't sell me a product that I wasn't allowed to modify and share my modifications back with others. The right to modify and redistribute transitively though the software license ga…

Nobody is laundering away proprietary livenses, because that code is not open source and not in public github repos. And OSS capabilities are now present in copilot, which is neither free nor open. Furthermore these contributions are making their way into proprietary code and the OSS licensing becomes even further watered down. This is the epitome of what copyleft is against!

I'm not sure this is true. Proprietary source code gets leaked and that can be used to train a NN. I find it likely that Copilot was trained against at least one non-OSS code base hosted on GitHub.

Second, if copyright is being laundered away we can get increasingly clever with how we liberate proprietary software. Today, decompiling and reverse engineering is a labor intensive process. That's the whole point of "open source" - that working in source is easier than working in bytecode. Given the hockey-stick of innovation happening in AI right now, I'd be surprised if we don't see AI assisted disassembly happening in the next decade. If you can go from bytecode to source code, that unlocks a lot. Even more so if you can go from bytecode to source code and feed that into a NN to liberate the code from its original license.

Re: We've filed a lawsuit against GitHub Copilot

#288

Post by MontagFTB »

elcomet wrote: post #267
vesinisa wrote: post #235

Earlier quoted context omitted.

Open source does not mean public domain. Open source specifically attaches limitations on how the code may be reused.

There are no limitations on reading the code to learn from it.

Perhaps the lawsuit contends that Copilot isn't in fact learning how to code, but is rather regurgitating information it has managed to glean and statistically categorize, without any real understanding as to what it was doing?

Re: We've filed a lawsuit against GitHub Copilot

#289

Post by tedunangst »

ilc wrote: post #132
bugfix-66 wrote: post #28

Ask HN: I want to modify the BSD 2-Clause Open Source License to explicitly prohibit the use of the licensed software in training systems like Microsoft's Copilot (and use during inference). How should the third clause be worded? The No-AI 3-Clause Open Source Software License Copyright (C) All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided tha…

If I read this right, I can't use auto-complete. No thanks.

Yeah, lol. New rule: code may be used for autocomplete, but only by a push down automata.

Re: We've filed a lawsuit against GitHub Copilot

#290

Post by 0cf8612b2e1e »

counttheforks wrote: post #105
Imnimo wrote: post #47

On page 18, they show Copilot produces the following code: >function isEven(n) { > return n % 2 === 0; >} They then say, "Copilot’s Output, like Codex’s, is derived from existing code. Namely, sample code that appears in the online book Mastering JS, written by Valeri Karpov." Surely everyone reading this has written that code verbatim at some point in their lives. How can they assert that this code is derived specif…

I wrote that exact function the other day, and I've never even heard of that book.

Should have used snake case. Would have avoided legal hot water and established precedent.
Post reply on HN