Laptop facial recognition defeated by Photoshop
Taking a long hard stare at biometric security
White hat security researchers have demoed how to bypass the facial recognition systems on several laptops.
The facial recognition software on Lenovo, Asus and Toshiba laptops (known as Veriface III, SmartLogon 1.0.0005 and Face Recognition 188.8.131.52, respectively) was compromised by security researchers including Duc Nguyen, senior researcher at Vietnamese security firm Bkis.
Details of the hack are were outlined by a presentation entitled Your face is NOT your Password during the Blackhat security conference in Washington earlier this week.
The laptops use webcams in conjunction with facial biometric software, as an alternative to more well-established login techniques. The researchers claim that the log-in approach can be defeated using nothing more sophisticated than a photograph of a PC's registered user, or even Photoshopped images.
Nguyen and his team created a large number of images to run what they described a "fake face bruteforce" attack to fool the systems, which in fairness are still in their infancy, into allowing a log-on. The approach can be compared to trying out a huge number of possible text passwords until the right combination is stumbled upon as part of a conventional brute-force dictionary attack.
Laptop makers ought to review the whole approach of facial recognition as a login technique, the researchers argue.
"Lenovo, Asus, and Toshiba are known as the first three big computer manufacturers to put that technology into practical use and to bring about greater convenience for their customers," Nguyen explains. "The one question to ask is whether such technology is really safe and secure for its users to enjoy."
"My research, which is concluded in this paper, will prove that the mechanisms used by those three vendors haven’t met the security requirements needed by an authentication system and that they cannot wholly protected their users from being tampered," he adds. ®
Sponsored: The Nuts and Bolts of Ransomware in 2016