Didn't get past the paywall but, recently a friend had their primary school targeted by blackmailers who took public photos of children and made (I'm told) extremely explicit images with them, threatening to release them if not paid. What's the defense? Intelligent screening of incoming messages so that the threat never reaches the blackmail target? I imagine they'll find an unprotected channel. Don't post innocuous…
In the US possession of CSAM is a strict-liability crime. If the models contain the data to produce it, the modelmakers should be liable.
Model makers, arguably, are like pencil manufacturers but in a World of good artists.
I think you can hold people to account for what they help to create, but not what they have potential to do.
Now, if models were trained on any csam, of course, the model's owners should be held you account.