Hmm. That's actually one area where Kagi can leverage the position of having accounts for users with expensive identities attached. It's going to be harder if one is running a search engine that's being anonymously used to deal with spam reports where someone is trying to get their competitors downranked or something.
I'm not at all sure that this can keep up with the rate of AI-generated spam websites alone. However, it might be sufficient to link human knowledge into the loop, build a training corpus to build an automated system that detects said spam websites.