A crowdfounded approach where readers flag content would probably be work well, without relying on AI. You'd possibly have to give more weight to readers who r

A crowdfounded approach where readers flag content would probably be work well, without relying on AI. You'd possibly have to give more weight to readers who reliably flag stuff, which, ironically, LLM users might be better at statistically.