Spam rules are usually separate from user filters or rules. By marking something as spam, the client will learn what is spam.
This should also be implemented at the server level. I’m actually not sure how MiaB has implemented its various spam protection features, but when I used to roll my own I recall this as being nearly default in SpamAssassin. When moving an email from the IMAP Inbox to the Junk folder, the user is training SpamAssassin on what is spam.