Wooh. This is startin' to move fast .
Not gonna lie, I think I glossed over or forgot about the handling of criminals but x) . That kind of overlaps with my back-and-forth with @joe_galindez here, but here's where I'm coming from: in a world where everyone has enough to eat, everyone has a place to live, everyone has access to cutting edge medicine, everyone is free to pursue their hobbies and passions, and there's a virtual reality wherein you can do even those things that you can't do in the real world for some reason or other, what manner of person commits any crime at all?
I would wager that only someone who just really, really wants to hurt someone else would do it, though I do suppose crimes of passion might still be a thing. I can't dictate how the AI would handle it, as that's outside the scope of the scenario. Unless someone wants to mandate that the answer is "all criminals are fired into the sun regardless of the offense," I'd assume the AI would do no better and no worse than human courts.
The scenario does not state that the AI is invested in preventing humans from reproducing, so we can assume that pregnant women will still give birth barring unforeseen misfortune. Also, I doubt an eternally-pregnant woman would be content. We could assume a Matrix scenario, but this isn't that, so that point is moot.
As for free will, it is outright stated at the start that people can do whatever they want. If you're making the point that a society is not a utopia if people aren't allowed to commit crimes (the common form this supposition takes, I've observed), I simply don't agree with that notion.
Also, the AI doesn't decide what is and isn't crime. Humans still had to program it, so humans still decided. Presumably a large council voted on it or something. The AI just carries out its programming. Whether or not that'll work out forever is unknowable, but we can assume based on the scenario presented that it wasn't programmed to consider something like poking someone a capital offense.