Calling the 50,000 networks a botnet is mischaracterizing what is going on here. The NSA only achieves its purpose when infecting a router or switch. This is what gives them access to all the data communicated on the attached network. Recall that with Belgacom the infection of IT staff computers was only an interim step, with the ultimate goal of infecting the GRX routers. A router does not run much of the software which makes botnets so useful to their controllers. The NSA would also not ever risk their surveillance capability by using control of a router for other purposes. If the router was not functioning well or doing very strange things then network IT staff are going to notice it and start investigating. Unless there was a stealthy root-kit (not an impossibility) on the router, the malware will be discovered and removed. The OS for routers has less of an attack surface than standard computer OSs. Even if Linux, or some other variation of UNIX is used then a lot of the capability, and thus attack surface, is disabled.

Once a router is infected, if a user’s computer or server was infected that malware isn’t so important anymore. Those, non-router, computers are updated much more frequently than routers or switches. Also, anti-virus software is not installed on routers. The NSA may even remove malware from non-routers to avoid detection. Then again, they may have achieved some very stealthy malware. I think it is less likely that arrangements are made with major AV companies to whitelist NSA malware. A whitelist is visible to too many people.

This particular leak is going to have an enormous impact on NSA capability. It would behoove any security executive for telecoms, or ISPs around the world to take a close look at their routers.

I count at least two countries who sent troops to fight and die in Afghanistan when the US bit off more than it could chew there.

This is going to stick in people’s memories come the next war.