That would certainly benefit companies developing generative AI. The sooner something loses copyright protection, the easier it is to use it as training data.