If you're opposed to machine learning making outputs based on training on copyrighted art, are you opposed to ai in general doing it, or opposed to the current types of models doing it?

That is, if you had C-3PO, would the same rule apply?