> SA performance varies greatly depending how well it has been tuned to the > users it protects.
What you say corresponds to my experience. Truth be told as well that it is a labour of love. And the resulting service is slow; the hardware is taxed to the point that I must either change hardware, which will consume more energy, or replace SA with something else. I read rspamd is a drop-in replacement. -------- Original Message -------- On Saturday, 05/30/26 at 19:32 Bill Cole <[email protected]> wrote: On 2026-05-30 at 00:07:47 UTC-0400 (Fri, 29 May 2026 21:07:47 -0700) Tom Williams via users <[email protected]> is rumored to have said: > Hi! A friend of mine is getting hammered with Costco, AAA, Lowe's > and similar spam. I don't know how they got her email address, but > she's getting flooded. Her email account is on a shared hosting > platform (Hostgator) and while Spamassassin is installed and > configured, it's blocking only so much of this. SA performance varies greatly depending how well it has been tuned to the users it protects. > Anyway, I'm writing because tonight, she started getting messages that > contain the text of a prompt for an AI engine to generate the spam > email. lol > > My questions: > > 1. Is having access to the prompt text useful in helping Spamassassin > detect and filter out AI generated spam? Unlikely, but I could be wrong. > 2. Would it be worthwhile to create any kind of rule to look for the > prompt text to help in filtering out the messages? Not having seen any such spam yet I can't say for sure, but I would think not. I doubt that any two spammers using LLMs would use the same prompt. It may be worth testing. > I have a few samples of these if anyone is interested. I suggest constructing rules and testing them. -- Bill Cole [email protected] or [email protected] (AKA @[email protected] and many *@billmail.scconsult.com addresses) Please keep discussion mailing list replies *on-list* Not Currently Available For Hire
