341–350 of 475 posts
I consider it a derivative work, it’s using statistical models to string tokens and has no ‘knowledge’ of a given block of code.
If the model is being trained on the code and is not copying and pasting it or including it directly from the various repositories then I would think a blanket attribution that covers all material used to train the model, basically a giant list of all of the authors, added to the Copilot repo should cover the attribution requirement. It'll be interesting to see how this plays out, to my understanding, these language…
Earlier quoted context omitted.
I just hope it doesn't end in Microsoft paying some (from their perspective) small fine that is just the cost of doing business.
The range of possible outcomes is enormous, I'll just wait by the sidelines but cherish the thought that moving out of GitHub when Microsoft bought it was the right decision. They can't be trusted, this has been proven over and over again and yet people keep falling for it. It's the fox guarding the chickens. I wrote about my misgivings at the time: https://jacquesmattheij.com/what-is-wrong-with-microsoft-buy...
> “Your honor, we needed so many works that it was simply not practical to ask permission of the creators.” I don’t find this argument convincing given the ability today to license many content types at scale for TDM, including images, music and yes, journal articles (See “Full disclosure” above), but it is an argument often offered by infringers. Why is this type of argument even valid? Isn't this fundamentally sayi…
> So it is not okay to infringe copyright at a small scale but okay to do it in a large scale? No, I think you're missing the "transformative" part. The line of argument isn't "we're going to resell millions of codebases as-is for pure profit", which would be undisputed copyright infringement. The argument is that something highly transformative (e.g. training models) isn't infringement at all, because transformative…
Earlier quoted context omitted.
The big difference there is that you never get a 1 to 1 copy of the source content out of the image models, where you often do with copilot. Whether your use is transformative is a part of the fair use legal test.
I seem quite a few exact copies of watermarks transormatively 'artistically generated' on AI images.
I don't understand this case. Is Copilot copying code wholesale and presenting it as its own? Because if not, then it doesn't need to attribute anything more than I need to attribute John Go or Edward PHP everytime I use a trick I picked up by reading their code. Obviously, if they trained Copilot on private code repos, that a whole other discussion. But I assume they didn't, so you don't even have the argument that…
Copyright covers expression, but not the ideas themselves. So it should be ok to mine ideas from projects, open or not, as long as the model doesn't reproduce expression. And even expression can be copied if it is small enough, trivial, public knowledge, the only obvious way to do something or an API call. If you want idea protection you need to look at patents.
No. You can't.
https://www.legalzoom.com/articles/can-you-patent-an-idea
'A machine sucking dust from a surface in order to clean it' is an idea.
Dyson design is an invention.
https://www.dyson.com/vacuum-cleaners/cordless/v15/detect/ye...
Here a broad description of differences between copyright and patent (trademark included):
https://copyrightalliance.org/faqs/difference-copyright-pate...
None of them covers the principle of idea.
And if you want an example of a dubious (in my opinion) attempt to patent an idea (patent troll):
https://patents.google.com/patent/EP3811791A1/en
While is it a patent trolling attempt (in my opinion) ?
Because it's basically an idea (adding a natural compound in ecigarette eliquids as sweetener) with an extensive list of layers covering general principles of everything related to ecigarette and ecigarette eliquid to pretend it's a specific formula (which is not) or an invention (which is neither because it's a natural compound known for its sweetener properties since 19th century).
I suspect that a lot of the people lining up on the other side against GitHub/Microsoft won't be so happy if the courts further lock down permissible uses under copyright across the board. More specifically, if Copilot breaks the "rules," so too does (probably--IANAL) pretty much every generative AI project out there. Restricting training to public domain datasets would be very limiting.
All copyrights and patents slow progress. If the desire for generative models is greater than the desire to hoover up cash, refine or end copyright and patent laws. Free Software people would build a statue to Microsoft if they started campaigning for an end to software copyrights.
Earlier quoted context omitted.
The service is displaying code... also I'm unaware of any TOS/EULA that cannot be amended or changed post agreement.
Not really about disallowing amendments, but at least sending out a notice of the changing terms. Like you get with your privacy policy. I'm pretty sure I didn't receive one about them using my public (although unpopular) open source code into their NN mixer. Edit: Anyway a bit outside the point. It being, when your ever expanding set of services incorporate your ownership in ways unforseen when the agreement was mad…