On Tue, 09 Jun 2026 15:54:15 +0200 Ludovic Courtès <[email protected]> wrote:
> Yes, what matters is authoring. (And as was brought up before, > “trivial” package definitions that can only be written in one way are > technically not copyrightable because they are not the result of > “human creative expression”.) About authoring, as I understand the above is valid for code written by humans and also for code that is generated by programs written by humans (though the logic for these is different and details matter a lot in that case). But for code generated by LLMs, I'm not a lawyer, and we were asked to wait for lawyers to come up with answers. And at least in the US, the copyright laws have a very long and bumpy history (this needs to be taken into account to in order to better predict the result of current/future lawsuits on LLMs), and the legal systems sometimes differ across jurisdictions. And as I understand, the legal systems also have their own logic, which non-lawyers tend to not know, and if I got that right, the consequence of all that is that the lawsuit results / interpretation of the law don't always match the common sense. So because of all that I think it would make more sense to not go too much in uncharted territory for now for for authoring and so just wait until we can have more clarity on the situation about authoring. Denis.
pgpppiW8VAiNA.pgp
Description: OpenPGP digital signature
