I admire your positive thinking, but it may also provide plausible deniability for legitimate CSAM, by your own logic. Either way, I see this being used to bully, blackmail, or worse. It’s not that we are going to stop AI development, nor that we should. Perhaps as it improves (remember, right now it’s the worst it will ever be), we can teach AI to recognize when it may be used for purposes like creating realistic CSAM or other such material, and have it log or report such uses.
I honestly don’t know the solution, but I don’t see the world ever getting “ho-hum, it’s all fake anyway” about minor involved pornography.