2018-11-16 23:45:06 2018-11-16 23:45:05 2018-11-16 23:45:05 1307709
Artificial AI 'Fingerprints' Can Fool Biometric Scanners
Despite their popularity, and a wide perception of offering strong and reliable security, biometric security technologies which measure things like fingerprints, faces, eyes, etc... are not without their vulnerabilities.
One argument against being overly concerned about biometric hacking, for everyday users, is that it would be too much trouble for most malicious actors to 'hack' the biometric data of average people. A fingerprint could be faked, for example, but the difficulty of doing so would make it worthwhile only against prime targets, such as the wealthy and powerful, whom presumably would also be likely to have additional forms of security to protect them.
Fingerprint scanners, in particular, are becoming a popular and highly touted security method, even becoming standard for high end phones and computing devices. That's why it is so disconcerting that a research team at New York University Tandon and Michigan State University have been able to create 'Skeleton Key' artificial fingerprints, called MasterPrints, able to fool them at alarming rates.
Using a type of AI tool called a Generative Adversarial Network (GAN), they were able to generate artificial fingerprints which contain features sufficiently common to many human fingerprints to be able to frequently trick the scanners. In their tests on a system with a supposed false match rate of 0.1%, these MasterPrints were accepted 22.5% of the time, or more than one time in five, rather than one time in a thousand.
This means that even average, everyday people who depend on fingerprint biometric security could become viable targets for various forms of hacking and identity theft, without the culprits ever needing their real fingerprint to do it (should nefarious agents gain access to similar technologies, of course).
You can add this to the long pile of reasons why I, personally, refuse to ever use any form of biometric security. They create an appearance of absolute security where none exists, and they're the only form of security that can be permanently hacked (since, after all, you can't change your fingerprints like a PIN Code).
#AI #ArtificialIntelligence #Biometrics #machinelearning #securitythreats #ai #biometrics #capacitivefingerprints #fingerprint #fingerprintscanner #fingerprintspoofing