Protecting copyright with AI

I’m still thinking about something I touched on in the OpenChain & Friends post: AI generated code isn’t copyrightable, because only human creations are eligible for copyright. In the past that used to be easy. Who if not a human programmer could have written the code? Today LLMs generate large parts of many code bases. And saying that I as a human reviewed every single line doesn’t count. The generative work, the concrete expression of the output is protected, not the after-the-fact approval...