Open source licenses cover derivative works. Machine learning based on pre-existing work should carry the downstream license. That is the responsibility of the human being initiating the machine learning algorithm, and you can't blame the machine. https://twitter.com/kellabyte/status/1411449622288076801
> I heard GitHub is training their co-pilot with other peoples copyrighted source code thus allowing copyrighted source code to be injected into other peoples products.

There was a time I felt GitHub was on the ethical side of things but that’s starting to fade.