Any blockning of "malverts" should arguably just emulate view so it costs the advertiser even more than the lost view. If Google don't want to do that themselves (which would be understandable) they could likely expose it in APIs so plugins like ad blockers can do it.
I don't think it would be fraud if they did it to ads from non-Google networks. But yeah, they shouldn't fake views on ads that come from their own networks.