Since herpes DOES infect humans, exposure to the virus would be the same risk either way. With some of the other vectors used, if it is something that wouldn't natively infect the species in which it is introduced, there is some risk if they are exposed to the "wild" virus. Care needs to be taken not to use them in environments where the wild virus is present to swap genetic material. The ones used in human trials require not just the missing genes from the virus, but also a second helper virus before they can replicate, so they're doubly protected from accidental transmission. Nonetheless, while there are people who think viral vectors could be used clinically, from where the current science stands, I don't see them as useful as much more than a delivery tool in the lab, under proper biosafety conditions where exposure to a wild virus, or exposure of any other animals to the infected ones are prevented. It doesn't mean I don't think they will ever be useful, but I think your concern is one that would need to be very carefully tested before going out and using this. And, even once a viral vector could be engineered that is completely safe, there would be a long struggle to get it accepted by the public as something they'd be willing to use. I think the initial reaction expressed in this thread is exactly the type of reaction most of the rest of the general public will have to such a therapy being introduced, and there will be a lot of objection to it at the outset.