

Everything posted on the public web is potential AI training data, federation is completely irrelevant to that.
The rest of your questions has the simple answer that a priori there is nothing “stopping” any of that. You should choose an instance whose admins are looking out for things like that and keeping your experience enjoyable, banning spambots or defederating from spambot farms when they are discovered.
First of all, can people stop empowering, rewarding and encouraging spammers by starting threads about them?
My understanding is that many people in the US and Canada identify with their ancestry? Lots of Americans think of themselves as “Irish” for example, some of them may actually be immigrants from Ireland, others may have more-or-less distant Irish ancestry, so this is what may be going on here.