Apparently chardet got Claude to rewrite the entire codebase from LGPL to MIT?

[dekkia@dekkia.com]

@Foxboron Considering that nobody can hold a copyright on AI-generated stuff, and therefore also can't release it under a different license, doesn't that mean this rewrite is basically public domain?

[duckattack@chaos.social]

@Foxboron lol, this is in a way what they suggest in this talk from #fosdem26: https://fosdem.org/2026/schedule/event/SUVS7G-lets_end_open_source_together_with_this_one_simple_trick/

[foxboron@chaos.social]

@dekkia
Public domain is not really a thing in most of the world. So "yes", for US. For EU it's more complicated.

[hannah@moonserver.yorha.nexus]

@Foxboron@chaos.social "Laundering code" through an LLM...

But:
Since the LLM-generated code cannot be copyrighted in any way, this entire project (or at least the part the LLM generated) is technically public domain.

Oh well. Not like certain entities care for the law! ​​

[dekkia@dekkia.com]

@Foxboron As far as I'm aware here in Germany it technically has copyright, but there's no owner who can enforce it.

I guess complicated is a good word for that.

[aparrish@friend.camp]

@Foxboron @dpk gross

[pixelcode@social.tchncs.de]

@dekkia @Foxboron Not really, a creator (in the copyright sense) can only be a human being, not a machine.

[brie@do.crimes.brie.gay]

@hannah@moonserver.yorha.nexus @Foxboron@chaos.social well, it would be public domain (by current rulings in the US) if the newer version is sufficiently different from the original LGPL to not be covered under that copyright

Very "funny" to license a repo as MIT when it is potentially either LGPL or public domain

[rootwyrm@weird.autos]

@Bubu @Foxboron somebody should inform PSF that in fact, chardet now has NO licensing and cannot be legally copyrighted or trademarked in any jurisdiction.

https://natlawreview.com/article/copyright-offices-latest-guidance-ai-and-copyrightability

https://fingfx.thomsonreuters.com/gfx/legaldocs/zdpxjnmmxpx/USPTO%20AI%20PATENTS%20squires.pdf