Can random mutations make HIV more lethal? Sure. The argument here is that lethality makes HIV "less fit" from an evolutionary standpoint, which makes sense, as, given the means of transmission, killing the host isn't a particularly useful thing for the virus to do.