eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Facial recognition technology may sound like a sexy, futuristic way to protect your laptop. But as security researcher Minh Duc Nguyen showed Feb. 19, the future is not always that bright when it comes to IT security.
Nguyen’s presentation was just one of several talks at Black Hat DC. The topics at the security conference ran the gamut from breaking biometrics to circumventing SSL (Secure Sockets Layer) encryption to thwarting forensics investigations when attacking Mac computers.
In the case of biometrics, Nguyen, a senior researcher at Bach Khoa Internetwork Security Center (Bkis), in Vietnam, found a way to bypass facial recognition technology used by laptop vendors Toshiba, Lenovo and Asustek Computer. The systems utilize a notebook computer’s built-in Webcam and facial recognition software to offer users a way to control access to their device other than passwords.
According to Nguyen, the systems can easily be defeated with digitized photos of the laptop’s owner.
“The problem is that the algorithm can not determine the difference between the images and the real face of the owner,” Nguyen explained to eWEEK after the presentation. “So, if an attacker has some pictures of the owner’s face, he can regenerate the fake face recognition suite[‘s] special pictures, and then he can log on to the laptops with these special pictures.”
He continued, “Our research showed that all of the three manufacturers have the vulnerabilities in their face authentication software. Even though they have applied more technical modifications to reduce the weakness, they have not been able to solve it completely. And we proved it in the Black Hat Conference. The result is this feature is not secure enough to protect the users[‘ computers] from being tampered [with].”
Nguyen recommended that Toshiba, Lenovo and Asus discontinue using the technology and warn users about its limitations.
Italian security researcher Vincenzo Iozzo’s advice for Apple Mac users was only slightly more optimistic. Iozzo presented to conference attendees a way to inject malicious code into Mac OS X memory without leaving any traces. In order to work, though, the attack has to be piggybacked on a reliable exploit for an unpatched OS X vulnerability, meaning IT administrators can stay ahead of the game by ensuring that computers are fully patched.
A hacker faces several challenges when he or she attacks a machine, Iozzo explained in an interview with eWEEK before the conference. Chief among them are avoiding detection, executing applications that might not be present on the attacked machine and not leaving traces on the hard disk, he said.
“My technique partially solves [these issues],” Iozzo said. “In fact, the whole attack is performed in-memory, which means that when the machine is powered off it isn’t possible to understand what happened because the attack leaves no traces on the machine. My technique allows an attacker to inject and execute binaries which are not present on the victim’s machine, so also the second problem is solved.
“Finally, when one wants to execute a binary into the victim’s machine it is necessary to execute a syscall, execve(). This might raise some alarms of IDS [intrusion detection system] systems or other types of security countermeasures and therefore detect the attacker,” he added.
When it comes to stealth, it’s no secret that there are tools out there designed to steal data and slip by unnoticed. A case in point is independent researcher Moxie Marlinspike’s SSLstrip. The tool enables users to launch a man-in-the-middle attack and silently change, intercept and log HTTPS traffic.
The net effect is that users can be tricked into thinking they are on a legitimate HTTPS site, such as an online banking log-in page, when in fact they are not. Though not completely invisible-an attentive victim may notice the allegedly secure page he or she is on is suddenly HTTP instead of HTTPS-the tool exploits a fundamental problem with the HTTPS protocol.
“The basic premise is that most people interact with HTTPS … through HTTP itself,” Marlinspike said after his presentation Feb. 18. “Most people come to SSL by either clicking on a link or getting redirected, and both of those are points that can be attacked … the basic thesis of the whole thing is that we have a secure protocol that basically depends on an insecure protocol.”
Those who missed Black Hat DC can look forward to Black Hat USA, which will be held July 25 to 30 in Las Vegas.