RE: https://mathstodon.xyz/@maxpool/117128107757895678
#IANAL , but from what I understand it is well-established in both German and American law that purely automated processes - and #AI / #LLM definitely qualify as such - cannot generate new copyrighted material. There _has_ to be human-created creative input.
The same applies to digital scans of #PublicDomain works - this is an automated process, and cannot turn a public domain work into something copyrighted. You actually have to do something _creative_ with the scan.
This is fairly important to my own work, since I translate German public domain texts into English, using digital scans of the texts as a base. Since I _am_ adding creative work to the originals, my translations _do_ have copyright.
Sure, I am immediately releasing them under a #CreativeCommons Zero license, which effectively makes them the equivalent of public domain again. But that's my _choice_ instead of something automatic.
