Posted Mar 25, 2009 at 23:55 by denisbider
Last edited Mar 26, 2009 at 00:07 by denisbider
Joanna Rutkowska
drives home the point that, even with the latest and greatest technologies that might protect us from all sorts of intentional and unintentional software and hardware backdoors - we still have to trust the CPU, which makes it exceedingly simple for the CIA to spy on you.
All they need is to make the CPU contain a backdoor such as this:
if (rax == MAGIC_1 && rcx == MAGIC_2) jmp [rbx]
This is:
- Trivial to hide among the 800 million gates of a modern processor.
- Exploitable in practically any program.
- Practically impossible to discover.
It gets worse: it doesn't even have to be the CIA. It can be any of the governments in the various countries where your CPU might have been manufactured.
It looks like, against the
most well-connected attackers, you can only consider yourself secure if you build
all your own hardware, and run
all your own software on it.
This post does not yet have any comments.