AI-generated 'Master' fingerprint can unlock most modern day smartphones

So you thought that the fingerprint sensor is one of the safest biometric methods to unlock your phone? The University of Michigan research department has developed an  AI-generated ‘master’ fingerprint which can unlock almost any modern day smartphone. [caption id=“attachment_5524841” align=“alignnone” width=“1024”]  The optical fingerprint scanner on the OnePlus 6T is slower and less reliable than the one it replaces. Image: Anirudh Regidi/Tech2[/caption] The paper submitted by the research time tiled DeepMasterPrints: Generating MasterPrints for Dictionary Attacks via Latent Variable Evolution, talks about a machine learning algorithm called “DeepMasterPrints”. This technology exploits two vulnerabilities in fingerprint-based authentication systems. The first flaw is that fingerprint scanners do not read the entire fingerprint. The second is that fingertip portions are more common than people are led to believe which means a partial fingerprint scan can be easily fooled. Using a neural network this team created artificial fingerprints that used “evolutionary optimization methods to find their best DeepMasterPrints”. The machine learning algorithm used was called “generative adversarial network” which created new fingerprints that matched other portions of the fingerprint. The artificially generated prints were accurate in replicating more than one in five real fingerprints in a database. “A similar setup to ours could be used for nefarious purposes, but it would likely not have the success rate we reported unless they optimized it for a smartphone system,” lead researcher Philip Bontrager of the New York University engineering school told  Gizmodo. However, if this hacker was to attack multiple fingerprint accessible accounts then the success rate will be much higher. Is face unlock now the safest way to keep your device locked?