It was always going to be a cat and mouse game because “AI” companies have decided to abandon ethics completely since there are few consequences when you they are just a shell company and the parent company keeps all of the resulting training data and money, so the company that does the training going bankrupt and abandoning responsibility is no issue. Sad that court system is so non-technical that they don’t see the training data produced by copyrighted material to be a copy of the material even if they were to decide that accessing the material was a violation.