I think a model that would make more sense is to punish bad behavior in the form of infringement, so if someone monetizes an AI output that infringes on someone

I think a model that would make more sense is to punish bad behavior in the form of infringement, so if someone monetizes an AI output that infringes on someone's copyright/trademark, then go after that person. Otherwise we are going to be completely stuck for the sake of some kind of outdated mentality around intellectual property.