I beg to differ. It might be a theoretical possibility, but one that is so remote as to not be worth thinking about. HIV already has a high rate of mutation, so the possibility is already out there. But you have to consider that for HIV to be successful, it can't kill the host too quickly or its opportunity to be passed on will be dramatically reduced. Even if the drug increased the error rate so that a "super" virus could be made, it would have also have a better chance of losing the mutations that gave it super status.
The threat is science falling woefully behind nature. Your point on killing the host too quickly is well taken by me yet already accounted for in nature. The weakest of those infected are the first to die of AID's and without viable treatment options the mortality rate would surely increase as would the spread of the virus by stronger carriers. Just a ugly form of natural selection. Inevitably science will stumble and nature will capitalize on our misfortune.
I agree what we are talking about is remote yet possible in our lifetime.