AI's been in use in commercial anti-spam for quite a while now - and on the flip side is also being used by the spam senders. Just another front on the unending war.

But spam (and phishing, and all malware) happens because humans get fooled by it. No reason to think AI will be any smarter.