Greg K-H
gregkh
My policy is that I do not take any output of any "AI" tools unless the providence of the data that was used to feed the AI tool can be proven to be under the proper copyright rules as to be compatible with the GPLv2 license.
So in other words, nothing from chatgpt at all, that's obviously full of copyrighted works that are not allowed to be reused in this manner.
6
67
98
🐧sima🐧
sima@chaos.social@gregkh I think a kernel community joint statement of this (like we have for the research shenanigans due to a big fail a while ago) would be really good. currently the AI patches are at the "mostly just entertaining, obvious nonsense" stage still, but I guess this could change quickly
maybe wrap it up together with the guidelines for dual-licensed code, since that's another topic I've seen pop up in a few places in public and private ...
0
1
3
Camille lives at Praxis Now
camille@social.praxis.nyc@gregkh
I don't think there's any way to enforce that, is there?
0
0
0
@gregkh how confident are you that you can detect violations of this policy?
1
0
0
Emil Velikov
xexaxo@fosstodon.org@craftyguy @gregkh I'm not sure if the maintainers should be in the role of actively policing. They are overloaded enough as-is already.
IMHO as abuses are brought to light - via research done by individuals, groups or otherwise - the work can be pulled out. Just like the unfortunate events 1-2 years ago.
1
0
0
1
0
0
0
0
0
Lars Marowsky-Brée 😷
larsmb@mastodon.online@gregkh Even if that provenance were proven, the output from an LLM can't be copyrighted itself, as far as I understand. So I'm not sure what that'd do to the kernel if it was a non-trivial contribution.
0
0
0
jzakotnik
jzakotnik@mastodon.social@gregkh maybe the idea of „copyright“ wasn’t so good after all.
0
0
0
Lars Marowsky-Brée 😷
larsmb@mastodon.online
0
0
0