Waking Up To Hardware Threats

For many years, hardware-based attacks were the thing of hypothetical conversations and security conference presentations.  2013 changed the nature of the game and as an industry we are waking up to the real threats posed by hardware.

First, we learned about the ability to weaken the random number generators in Intel CPUs during the manufacturing process in a manner that is extremely difficult to detect.  Then, we had a report about a researcher creating malware resident on a video card – of course, the story there was that the researcher created a prototype anti-malware tool to detect malware-laden video cards.  While these were interesting stories, they were essentially more of the same – theoretical attacks.

Then later in the year, Dragos Ruiu began discussing what he believed to be a potent new piece of malware that he named “bad bios“.  This created a firestorm of speculation, both that Dragos is crazy and that some government has it out for him.  A lot was written about why this was and was not possible.   Bad bios seemed to represent the worst case scenario for malware – very hard to detect, persistent across operating system re-installations and able to communicate across air gaps.  All with no indication of what the intended purpose of the malware is, if the malware if even real.

Most recently, we learned from leaked NSA documents reported by Der Spiegel that the NSA will intercept shipments of computers destined for target individuals or organizations and “…often seek to place their malicious code in BIOS, software located directly on a computer’s motherboard…” and “… also attack firmware on computer hard drives…” with “…spyware capable of embedding itself unnoticed into hard drives…” We appear to have jumped from the realm of hypothetical and theoretical attacks involving hardware into a world where this is apparently a commonplace and well established practice.

At the same time, we have seen a hardware hacker take control of a Western Digital hard drive and essentially install Linux on the embedded controller, theorizing that such a strategy would work to hide persistent malware or even destroy data on a disk that is being copied, but otherwise allowing normal access to data contained on the drive.

As well, at the annual CCC in Germany, a presentation was delivered on the ability to take control of the embedded microcontroller of SD cards, similarly offering the ability to hide malware or data.

Defending against hardware based attacks is going to be very challenging.  I see a lot of opportunity for security companies to create strategies to attest to the integrity of attached devices, like hard drives, BIOS and SD cards – not just the contents, but the actual controllers, if such a thing is even realistic to accomplish.

This is an interesting new world that we are waking up to, and I look forward to seeing how our industry will take on the challenges it presents.