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 acc

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?