Given that we now have PhD-level agentic reasoning AI models I was wondering how is it possible that spam still is a problem.

Does spam filtering requires super intelligence?