In this install we use Herald to add a "blocking reviewer" to ensure code is reviewed by an appropriate upstream representative.

My question to you would by why are users who are not assigned to a Diff reviewing and accepting it, and if they are, why are authors then taking that as actually being accepted. If this isn't the case and your asking based on theoretical abuse, I'd suggest waiting and seeing if it's actually a problem. Because of Phabricator's auditability, it's generally expected users will do the correct thing.