• 39 Posts
  • 779 Comments
Joined 10 months ago
cake
Cake day: September 1st, 2023

help-circle
























  • I don’t know if a verdict has already been on the topic, nor by whom, but it’s my impression that the issue of copyrighted material used to train LLMs or GPTs is still hotly debated. IIRC Microsoft was sued because of Copilot on the grounds of copyright infringement. Especially with stuff like GPLv3 where derivative works must be of the same licence.

    Personally, it’s not much of an issue to me whether my public musings will be scraped. I’d rather they are scraped and used for opensource stuff, but there’s no way for me to enforce that. However, if a model starts spitting out “CC BY-NC-SA 4.0”, I’d at least find that funny and maybe it might even help bring a case on the grounds of breaching the “Non Commercial” clause of the licence.