Ask HN: Is it possible that hardware backdoors have direct access?

10 points by wyck ↗ HN
Is it a possibility the PRISM related direct access is a hardware backdoor?

References:

NSA has their own chip manufacturer: http://en.wikipedia.org/wiki/National_Security_Agency#Role_in_scientific_research_and_development

Hardware backdoors are practical (defcon 2012 slideshow): http://www.slideshare.net/endrazine/defcon-hardware-backdooring-is-practical

Backdoor in military chips from China: http://www.scribd.com/doc/95282643/Backdoors-Embedded-in-DoD-Microchips-From-China

Rakshasa: http://www.extremetech.com/computing/133773-rakshasa-the-hardware-backdoor-that-china-could-embed-in-every-computer

2 comments

[ 4.4 ms ] story [ 12.9 ms ] thread
I'm only throwing my opinion, I'm not really into security but I spend some time studying networks and low level stuff as a hobby, any specialist will probably point my errors below.

Depends on the device no? Suppose you have a chip with a backdoor, if said chip depends on a OS to operate and you can for example block connections (using a packet filter, either blacklisting the undesirable ones or drop all except the whitelisted ones), then whatever backdoor is there will need an unblocked connection to be activated.

As I see it the majority of hardware backdoors would only be useful if you can have some sort of access to the device, either remotely in a network the device is connected, physically (as far as I understand that's how Stuxnet spread in Iran) or if the user willingly executes some code that exposes the backdoor to you over some routing.

If it was a hardware backdoor, why would it be limited to the specific subset of companies it claims to be limited to? Occam's Razor strongly suggests it's just poor wording on the part of whatever schlub wrote the slide.