Interesting, was going to say basically the same thing. At what point do we start designing RNA with a delivery mechanism that attaches itself to a specific bacterial type and eliminates it?
Consider 'fly by wire' as an analogy. There was great emotion (and still is) around taking the pilot of an aircraft out of direct control of that aircraft. First required by dynamically unstable aircraft (the F-117) this is now a standard feature on large passenger aircraft. The problems are still the same though, one bad line of code (or perhaps a few bad lines) and the plane flies into the ground killing everyone.
Similarly with engineered RNA. We actually do this a lot these days (very interesting bio hacking going on) and with DNA printers you can pretty much design your own bug, print it out, load it into a waiting husk of an E. Coli bacterium and boom, new bug.) Understanding isn't quite there yet but it will be. And knowing the risks of a particular DNA pattern will be just as understandable as a knowing the risks of a piece of flight software.
[1] http://www.3ders.org/articles/20130324-synthesizing-dna-time...
(I think your analogy hides the potential impact of a mutating virus. Fly by wire only puts air travel at risk.)