[lumost]

These are so absolutely ridiculous. The only (dumb) reason I can think of is that there is a large ML model which got used to seeing certain character sequences as not spam and spammers are starting to exploit it.

It's kinda shocking as Gmail spam filtering was virtually flawless for over a decade, and now it's falling apart.

[Maken]

Maybe the spammers learned about adversarial networks?

[lumost]

Any non-ML method would trivially catch these spammers though. They share the common traits of

1. Sending identical emails to large numbers of users. 2. Using fake emails from fake domains. 3. Not looking at all like a regular email.

This is the stuff that 90s era spam filters could block.

[hulitu]

> This is the stuff that 90s era spam filters could block.

But we are now in 2022. Those filters are loooong gone. I bet that in 20 years they will be sold as new technology.