All public GitHub code was used in training Copilot • Hacker News Board

Re: All public GitHub code was used in training Copilot

#201

Post by shakna » Thu, Jul 08, 2021, 3:33 PM UTC

AaronFriel wrote: post #193

keonix wrote: post #185

If the training set contains verbatim (A)GPL code does this mean that Copilot also should be distributed by Microsoft under GPL? Because without it Copilot (as it is distributed by Microsoft) couldn't be built, wouldn't it make it a derivative work of GPL'd code (and obviously every other license)? I see a lot of people comparing human learning to machine learning in the comments, but there is a huge difference - we…

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

> No, see Authors Guild v. Google.

That case required that the output be transformative, in that "words in books are being used in a way they have not been used before".

Copilot only fits the transformative aspect if it is not directly reciting code, that already exists in the form that it is redistributing. So long as it does so, it fails to meet the criteria.

Re: All public GitHub code was used in training Copilot

#202

Post by _Understated_ » Thu, Jul 08, 2021, 3:34 PM UTC

AaronFriel wrote: post #193

keonix wrote: post #185

If the training set contains verbatim (A)GPL code does this mean that Copilot also should be distributed by Microsoft under GPL? Because without it Copilot (as it is distributed by Microsoft) couldn't be built, wouldn't it make it a derivative work of GPL'd code (and obviously every other license)? I see a lot of people comparing human learning to machine learning in the comments, but there is a huge difference - we…

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

> By comparison, Copilot is even more obviously fair use.

Not sure I see it that way.

If I take your hard work that you clearly marked with a GPL license and then make money from it, not quite directly, but very closely, how is that fair use? Or legal?

Copying and storing a book isn't recreating another book from it. Copilot is creating new stuff from the contents of the "books" in this case.

Edit: I misunderstood fair use as it turns out...

Re: All public GitHub code was used in training Copilot

#203

Post by blibble » Thu, Jul 08, 2021, 3:34 PM UTC

swiley wrote: post #194

agilob wrote: post #134

This means that also all illegally leaked codes from Apple, CDPR, Intel, NSA and Microsoft leaks are used in the models? iBoot, Witcher 3? Gwent? NSA backdoors? Does the copilot still learn from new repos? Can I post github enterprise code publicly to let it learn from it? Serious answers only please

Has Microsoft just killed source code copyright? That would definitely be a win.

it would be a win for Microsoft that don't distribute their source code

whereas for open source it's a disaster

Re: All public GitHub code was used in training Copilot

#204

Post by echelon » Thu, Jul 08, 2021, 3:35 PM UTC

AaronFriel wrote: post #193

keonix wrote: post #185

If the training set contains verbatim (A)GPL code does this mean that Copilot also should be distributed by Microsoft under GPL? Because without it Copilot (as it is distributed by Microsoft) couldn't be built, wouldn't it make it a derivative work of GPL'd code (and obviously every other license)? I see a lot of people comparing human learning to machine learning in the comments, but there is a huge difference - we…

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

> Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books.

For commercial use and derivative works?

Authors won't incorporate snippets of books into new works unless they're reviews. Copilot is different.

Re: All public GitHub code was used in training Copilot

#205

Post by ghoward » Thu, Jul 08, 2021, 3:35 PM UTC

nradov wrote: post #177

p0ckets wrote: post #148

Earlier quoted context omitted.

I think to actually poison the well, we should add code to existing repos with dead code clearly labelled as "the way that things shouldn't be done" that are wrong in subtle ways. So every time we fix a security issue, we keep the version with the bug with some comments indicating what's wrong with it. Of course, this only works until the AI is trained to weigh the code based on how often the code is called.

The notion of intentionally polluting and over complicating your code base just to "poison the well" is bizarre. Talk about cutting off your nose to spite your face. If you don't want others to use your code then the solution is very simple. Keep it on a secure private server and don't publicly release it.

Keeping it private is one option, but I really want my end users to have the freedom to modify the code for themselves.

Re: All public GitHub code was used in training Copilot

#206

Post by jcelerier » Thu, Jul 08, 2021, 3:35 PM UTC

laurowyn wrote: post #129

Could this be the beginning of the true test of open source licenses? My understanding is that there has never been a ruling by a court to give precedence to the validity or scope of any open source license. I can see a class action suit coming on behalf of all GPL licensed code authors.

What ? There have been plenty of GPL cases defended in court.

https://en.m.wikipedia.org/wiki/Open_source_license_litigati...

Re: All public GitHub code was used in training Copilot

#207

Post by echelon » Thu, Jul 08, 2021, 3:36 PM UTC

swiley wrote: post #194

agilob wrote: post #134

This means that also all illegally leaked codes from Apple, CDPR, Intel, NSA and Microsoft leaks are used in the models? iBoot, Witcher 3? Gwent? NSA backdoors? Does the copilot still learn from new repos? Can I post github enterprise code publicly to let it learn from it? Serious answers only please

Has Microsoft just killed source code copyright? That would definitely be a win.

But who says the code has to be available to anyone but Microsoft?

Remember that Amazon won off the back of open source. Now all the open source servers and databases are Amazon products.

Re: All public GitHub code was used in training Copilot

#208

Post by Hamuko » Thu, Jul 08, 2021, 3:36 PM UTC

_Understated_ wrote: post #202

AaronFriel wrote: post #193

Earlier quoted context omitted.

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

> By comparison, Copilot is even more obviously fair use. Not sure I see it that way. If I take your hard work that you clearly marked with a GPL license and then make money from it, not quite directly, but very closely, how is that fair use? Or legal? Copying and storing a book isn't recreating another book from it. Copilot is creating new stuff from the contents of the "books" in this case. Edit: I misunderstood fa…

>If I take your hard work that you clearly marked with a GPL license and then make money from it, not quite directly, but very closely, how is that fair use? Or legal?

You can wipe your ass with the GPL license if your use of the product falls within Fair Use.

You can actually take snippets from commercial movies and post them onto YouTube if your YouTube video is transformative enough for your usage to be considered fair use. Well, theoretically at least - in reality YouTube might automatically copyright strike it.

>Copying and storing a book isn't recreating another book from it.

That doesn't mean that GitHub has to redistribute Copilot under GPL. However, the end user could potentially have to if they use Copilot to generate new code that happens to copy GPL code verbatim.

Re: All public GitHub code was used in training Copilot

#209

Post by IgorPartola » Thu, Jul 08, 2021, 3:37 PM UTC

AaronFriel wrote: post #193

keonix wrote: post #185

If the training set contains verbatim (A)GPL code does this mean that Copilot also should be distributed by Microsoft under GPL? Because without it Copilot (as it is distributed by Microsoft) couldn't be built, wouldn't it make it a derivative work of GPL'd code (and obviously every other license)? I see a lot of people comparing human learning to machine learning in the comments, but there is a huge difference - we…

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

I think this is the correct answer. IANAL but the copilot code vs the copilot training data are different things and licensing for one shouldn’t affect the other, right? And the fact that training data happens to also be code is incidental.

Re: All public GitHub code was used in training Copilot

#210

Post by IgorPartola » Thu, Jul 08, 2021, 3:39 PM UTC

_Understated_ wrote: post #202

AaronFriel wrote: post #193

Earlier quoted context omitted.

No, see Authors Guild v. Google. Even without a license or permission, fair use permits the mass scanning of books, the storage of the content of those books, and rendering verbatim snippets of those books. The Google Books site is not a derivative work of the millions of authors they copied from, and if they did copy any coincidentally GPL, AGPL, or creative commons copyleft work, the fair use exception applies befo…

> By comparison, Copilot is even more obviously fair use. Not sure I see it that way. If I take your hard work that you clearly marked with a GPL license and then make money from it, not quite directly, but very closely, how is that fair use? Or legal? Copying and storing a book isn't recreating another book from it. Copilot is creating new stuff from the contents of the "books" in this case. Edit: I misunderstood fa…

If you view GPL code with your browser would that mean that your browser now has to be GPL as well? In the sense that copilot is not much different than a browser for Stack Overflow with some automation, why would it need to be GPLed? Your own code on the other hand…