I do receive phishing emails; some of them are so well-crafted that I'm sure they've fooled some people out there. To the point where I've created a folder called "nice_try_phishing" where I collect them for further investigation. For example, one email was sent before I renewed a domain as a reminder to renew, with legitimate links to the domain registrar except for the action link. They had the registrar's domain name too, but with a different, very similar TLD. Another one is a "failed email delivery," and they did the research about which service I'm using to mimic such an automated message, with loaded links.

Whether they are AI or not, I have no idea, but sometimes, and recently in emails, I purposely make a typo or grammar mistake to add some "human" touch to it, knowing that an AI will always type a perfect one.