The humble mosquito, and the deadly diseases it carries, is estimated to have been responsible for as many as 46 billion deaths over the history of our species. That staggering number is even more frightening in context - it means that mosquitoes are alleged to have killed more than half the humans that ever lived.
In short, the modified genes affect only the female mosquitoes, rendering them flightless. The larvae hatch on the water, and the females are unable to leave, rendering them harmless to humans and leaving them to die. The males are unaffected, so they mature normally, then mate with other females to pass the genetic modification on.
I've often wondered why stuff like this hasn't been done before. Admittedly I'm not a geneticist, but it would seem we have gotten pretty good at genetically engineering simple lifeforms. Particularly in this case, since we don't care much about the overall health of the insect. Just produce insects that carry a gene that makes them sterile when there are two of them, and release tons of them. It would seem like this has been done, although using radiation to sterilize them as opposed to genetics.
A few years ago, I read something about genes that could affect the odds that they would appear in the sex cells. A sex cell is supposed to have a random gene picked from the two possible genes (one from your father one from your mother). These genes would increase the odds that they would be picked over the other choice to something over 50%. This struck me as an interesting way to genetically engineer some sort of species as a weapon that dies off, (think the xenomorph from the Alien movies). You create perfectly normal aliens and unleash them on some world. However, included in the initial group would be a few individuals with a certain gene. This gene would have various effects based on if the individual had two copies or one, and if they were male or female. A male with one copy would be more likely to pass on the gene (maybe 90%). A male with two copies would be sterile. It would have no effect on females, they'd just carry it normally.
I wondered how effective this gene would be, and did some spreadsheet calculations. I then wrote a basic program to simulate the odds. I discovered that the gene was super effective as long as it had high enough probability. The magic threshold appeared to be 86%. At 85% the population would stabilize with a small core group of gene free males. At 86% it would always wipe out the population, even when introduced into a large stable population. I'm sure there's some interesting mathematical explanation for that point, but I don't know what it is, and I didn't really want to write a whole detailed post about this.
Oddly, I never added it to my site. I vaguely remember not considering it done, but maybe that was before I realized that anything I do is done the first time I stop working on it. I added it and it's there now, at the end of the C++ section.