No, they wouldn't.

We've been doing Bayesian content (aka spam) filtering for over 20 years, based in no small part on Paul Graham's essay "A plan for spam". According to HP [1], a home computer at the time had a single 1.5Ghz core and 256Mb of RAM.

Using LLMs would achieve essentially the same while requiring a couple orders of magnitude more resources.

[1] https://www.hp.com/us-en/shop/tech-takes/specifications-pers...

Would those filters be keyword-based only? One benefit of an LLM-based filter I can imagine is that it has a much better understanding of the meaning of text.