The final approval settles one case, but it doesn't resolve the broader issue of using copyrighted works to train AI models.
