I personally found it really amusing how they weaponized the legal system to DMCA all the claude code source code repositories. Code ingested into the model is not copyrightable, but produced code apparently is when by legal definition computer generated code can not be copyrighted and that's one of their primary arguments in legal cases.
I did an interesting spin to avoid it: generated Claude abstracted as a set of PRD style architectural instructions so you can do your claude-code inspired program effortlessly.
There's even a GUI called claudia for a piecemeal extraction with a PRD.
I wonder if GitHub would rule it a copyright violation if the source code was rewritten by an agent, i.e. copy my answers but change a few words. Legally, if the original source code is copyrighted then an agent rewriting it likely doesn't lose that copyright, but I wonder if GitHub would go through the effort of determining whether it was a derived work.
I guess you'd be assuming that both the original CC code and its ports are computer/AI-generated? As a lawyer, though, I'd still maintain that you wouldn't need much original human input in the CC code to kind of ruin that theory. The threshold for copyright protection isn't that high, really.
The copyright office has indicated that AI generated elements lacking human contributions to the expression (rather than input) would still not be copyrightable even if some human authored elements are made to the expression as well.
Seems like it would be a nightmare to provide evidence of what parts of a half a million line codebase were written by humans if no one bothered to track it.
