Comment by PaulDavisThe1st
5 months ago
I am confused. In the USA, there has been a clear rule that machine-generated code cannot be copyright. If the "new implementation" was in fact created by Claude (which is my impression), then nobody holds any copyright to the code, and it cannot be licensed under any license at all.
I am sure I am missing something ... what is it?
This is effectively a contract. You can put anything you want in a contract, but contracts are enforceable to only to the extent they comply with the law (statutes, case law, the constitution, etc.)
So to settle this, someone needs to violate this license and get sued. Or maybe proactively sue?
Remember: this applies to all LLM-generated code, not just chardet. No LLM-generated code is copyrightable, and thus cannot be licensed. The legal challenge could come in any context where LLMs have been used and the code placed under any license (proprietary or otherwise).
Which is going cause a collision between the "not copyrightable" and "derived from copyrighted work" angles.
The machine cannot get authorship, but just as images created by humans with PhotoShop and all kinds of machinery are still copyrighted to the human creator - unless they explicitly set up some circumstances where the process of creation happens completely without them - code/software produced by a machine instructed by a human should get copyright (either original or derivative).
Unless the human is so far removed from the output. (And how far is far enough is probably very much depends on the circumstances and unless/until case law or Congress gives us some unifying criteria, it's going to be up to how the judge and the jury feels.)
..
For example someone set up a system where their dog ends up promoting some AI to make video games. This might be the closest to the case of that photo.
Though there the court ruled only that PETA (as friend of the monkey) cannot sue the photographer, because the monkey cannot be a copyright holder, but very importantly it didn't rule on the authorship of the photographer. (And thus the Wikimedia metadata which states that the image is in thr public domain, is simply their opinion.)