Don’t most licenses require at least attribution? I don’t believe GitHub is restricting themselves to only licenses that don’t. In fact the only software licenses I can think of that don’t require attribution are 0BSD, WTFPL, CC0, MIT-0 and Unlicense, and they all aren’t super popular. Also in some countries creators have inalienable moral rights which can be enforced regardless of the license. For example in Germany it is impossible to relinquish certain rights you have as the creator of a work, including the right to attribution.

This is an important and overlooked point. Even common permissive licenses (ISC / MIT / Apache-2.0) require attribution

Just as a mind experiment: couldn't CoPilot just publish a list of every github user and attribute the work to all of them?

CoPilot is a black box at the moment. Microsoft claims they used the public corpus on GitHub. There are plenty of GPL, AGPL, and "source available" projects in the public corpus. So what exactly is the licensing?

The argument may make sense if they limited themselves to public-domain (CC0) works, but that is not what happened here. If CoPilot attributed something to an AGPL project, does it mean the "virality" applies to all projects that use code from CoPilot?

There's also a good amount of commercial and leaked source code on GitHub, including MS's own leaked Windows XP source. I haven't played around with Copilot yet, but if I ever do I plan on copy/pasting some win32 API definitions to see if I can get it to spit out any of the leaked source.

> if I ever do I plan on copy/pasting some win32 API definitions to see if I can get it to spit out any of the leaked source.

If that works, then I can't wait for that to be a boon for Wine and ReactOS: "Microsoft itself provided this code and allowed us to use it, so therefore it's totally legal. Neener neener."

Some trivia: CC0 is a public domain declaration.. at least in the US. There is no process by which an author can make their works public domain, CC0 is just a (weak) promise that the copyright holder will treat the work as if it were public domain.

This feels like a tool that can easily be destroyed by a lawsuit, I can't imagine a TOS can force you to give away your copy rights (especially if they allow and encourage you to post your own copyright).

If it can't then Wikipedia is doomed; its entire licensing status rests on the notion that editors grant such a license as part of their clickwrap ToS.

Does GitHub verify that the code that is in my repository is actually in accordance to the license that I’ve added? I could just upload any proprietary code with an incorrect license, and GitHub would just use that to feed their AI. Like any other dependency that you incorporate into your application, GitHub should verify/audit whether the license allows them to do so.

> If your code is using a license that allows it, how could you possibly opt-out aside from using a different license?

A repo setting that instructs Github not to use your code for Copilot, it could be a similar option as turning Discussions on / off.

If they really want to win developers over they would even have Copilot scanning disabled by default but that'll never happen.

Even if Github did provide that setting, as a courtesy, someone could clone / fork the code to another repo (if you use any licence that allows it) and not enable that setting.

Sure that's possible, but there's a huuuge difference between Possible and Default Behavior.

In a case like this, GitHub itself could set up a bot account that forks all projects as soon as you make the switch. The company in fact would be incentivized to do so.

Sounds like you want a new license that just prohibits use by one company for one purpose.

There are other AI-based code completion systems than Copilot, at least Tabnine[1] and Kite[2] come to mind, I'm sure there are more.

[1] https://www.tabnine.com/

[2] https://www.kite.com/

As of today there is a new one...

"Now in Preview – Amazon CodeWhisperer"

https://aws.amazon.com/blogs/aws/now-in-preview-amazon-codew...

is there something wrong with that?

Not at all, you can put any license on your code that you want.

I'm not sure using a different license actually opts you out. By merely hosting your code on GitHub you grant them the right to analyze your code on their servers[1]

They may be morally in the wrong, but I'm unsure they are legally in the wrong here. To boot, denying them the right to create this tool in your license is technically a violation of OSS principles and problematic

[1]: https://docs.github.com/en/site-policy/github-terms/github-t...

> This license does not grant GitHub the right to sell Your Content. It also does not grant GitHub the right to otherwise distribute or use Your Content outside of our provision of the Service, except that as part of the right to archive Your Content, GitHub may permit our partners to store and archive Your Content in public repositories in connection with the GitHub Arctic Code Vault and GitHub Archive Program.

Wouldn't this be a violation?

Microsoft could provide an opt-out for projects or even contributors, regardless of licence.