Keyloggers Aren't Viruses... Are They?

5:38 PM -- The SANS Internet Storm Center (ISC) is a great resource for things that go bump in the wire. For instance, there are right now posts about current vulnerabilities in Linux kernels 2.6.24.1 and prior, QuickTime, Adobe Reader, and Firefox. Since the ISC incident handlers represent countries from all over the world, they receive and analyze reports of attacks and exploits before most of us get to see them on our network (unless of course, you work for a large university).

On rare occasion, I don't see eye to eye with one of their posts, such as the recent one titled “Does your anti-virus detect old keyloggers?” Keyloggers are not viruses. I'm not the only one that thinks that, am I? Just because malware authors and attackers include keylogging within their malware or install a keylogger to catch passwords, doesn’t mean that they should be flagged as a virus. I find that line of reasoning as annoying as having antivirus catch copies of nmap or similar tools on my systems.

Out of the more than 30 virus scanners VirusTotal uses to scan suspicious files, only four found the Tiny Keylogger file to be suspicious, with three of them flagging it as a keylogger and the other simply calling it Spyware.Gen (thanks, eSafe, for the useless description).

One ISC reader posted a comment that McAfee only detects keyloggers if detection for “potentially unwanted applications” is enabled. I suspect that this is the case for most antivirus software, but I don’t think that keylogging detection should be a function of an antivirus signature -- it should be behavioral.

Catching a “known bad file” takes more work because all known bad files have to be known, and they must have signatures created for them. But a keylogger should be detected since it displays suspicious behavior, such as hooking into the underlying operating system to intercept key presses from the keyboard.

I guess it boils down to what you want from an antivirus product. For me, I want it to do signature detection well -- but behavioral detection, better. Others might have a different opinion, but it goes to show that when you’re testing files for suspicious behavior, you can’t rely on just one source for detection, or even one method of detection.

– John H. Sawyer is a security geek on the IT Security Team at the University of Florida. He enjoys taking long war walks on the beach and riding pwnies. When he's not fighting flaming, malware-infested machines or performing autopsies on blitzed boxes, he can usually be found hanging with his family, bouncing a baby on one knee and balancing a laptop on the other. Special to Dark Reading