Regurgitating copyrighted material for profit is a concern. But I fail to understand why training on copyrighted material is a problem. Have we not all trained our brains reading/listening copyrighted material? Then why it is wrong for AI to do the same?