News for hackers. Stuff that matters.
Re:Google took three months to remove scam app that stole ov... by throwaway2037 on Monday August 19, 2024 @01:36AM (#41288102)
To be clear, I assume that modern, sophisticated spam operations are "leading the league" in LLM usage. It must be much harder to stop spam when each email can be individualized by an LLM. And let's be real: LLMs are already very, very good at producing text that sounds believable. I, myself, have been fooled many times already by recent spam, that is so much more believable than two year ago.

If GMail is getting worse, I can imagine that other, smaller mail services are getting much worse. The best explanation that I have read about why Google (and other major providers) are so good at spam filtering: They can observe a huge portion of the world's email, so they have the best training sets.

It is interesting that we never hear from GMail folks on HN. You see all kinds of Googlers pop-up into discussions with interesting insights about how the sausage is made. However, I cannot recall anyone from GMail appearing on HN to share some interesting behind-the-scenes stories.

Re:Google took three months to remove scam app that stole ov... by whatshisface on Monday August 19, 2024 @01:47AM (#41288146)
SpamAssassin seems to work for me.
Re:Google took three months to remove scam app that stole ov... by kuschku on Tuesday August 20, 2024 @08:54AM (#41299630)
> The problem with building a bear-safe trashcan is that "There is considerable overlap between the intelligence of the smartest bears and the dumbest tourists"

Google can't create a good spam filter, because enough users actually want to receive spam.

A machine can't distinguish between male enhancement pill spam and a newsletter recommending alex jones' supplements.

Or between "join our MLM today" and emails from recruiters.

Or between sales people sending cold emails and social engineering.

Google should ban both, but if they did, there'd be a huge outrage from idiots that actually want to receive content that's basically indistinguishable.

When pre-Musk Twitter tried to reduce the wave of fake news, they actually had to explicitly whitelist republican accounts because

> Most fake news on Twitter is spread by […] Republican, middle-aged White women residing in three conservative states, Arizona, Florida, and Texas