I think I understand the problem you are pointing at. For example, children porn via AI? My question that is more epistemological and take aside ethics is if you can stop AI to produce those images when all the tools are available. I can compare this (not in ethics terms but in feasibility) to the US government push to control the information about how to make bombs or cryptography in the 90s. They couldn't but they can prosecute people who take advantage of this technologies to commit a crime. I don't know how this applies to people watching children porn that is produced by AI. Here I understand this is an ethics question that talks about the person consuming this and not the technology.