This is happening today.

The answer is trusted relays and Web of Trust

Long-time nostr user. My feeds are all spam free. Not to say there aren't any other problems :-)

I guess it's easier with commercial spam when they have to advertise something, and that could be detected (website, product, etc)

But when it's just a new account with a few LLM-generated replies - how would trusted relay automatically detect the malicious intent?