Sunday , January 24 2021

Artificial intelligence can mimic human fingerprints.




Photo: mobile phone with fingerprint scanning capabilities.

Photo: mobile phone with fingerprint scanning capabilities. |

Oatawa / Shutterstock

The fingerprint scanner is currently one of the important security features in a mobile phone. This feature provides security, because everyone has their own fingerprint.

Recently, however, artificial intelligence (AI), which is becoming more complex, can actually mimic the security system by scanning the finger and retina. reportedly EngadgetOn Friday (11/16/2018) the AI ​​function was developed by a group of researchers from New York University, USA (USA).

The researchers said that an AI, called DeepMasterPrints, can make replicas of more than five fingerprints with an accuracy of 23 percent. Of course, this percentage is very large.

Because the comparison of matches between one fingerprint with another fingerprint can reach one in a thousand. However, when the compatibility level reaches one in a hundred, DeepMasterPrints is able to simulate accuracy up to 77 percent.

Study leader Philip Bontrager told The keeperOn Thursday (11/15) the developed method can be widely applied. Not only fingerprints, but also fingerprints. By the way, the researchers used a number of aspects of the fingerprint scanner, one of which was ergonomic.

Most fingerprint scanners do not immediately read the entire finger. They store any part that touches the surface of the scanner.

That's why you need to lift, lower your finger and rotate it when installing the TouchID on an iOS device or opening a fingerprint on an Android device.

The method of work does not include certain parts in order to compare it with a full human fingerprint. The scanner compares only one part with another.

This fingerprint can be very effective for passing through a system with many fingerprints stored in it. A criminal may have a better chance of success through court and mistake – similar to how hackers conduct attacks brute force or hack password site.

Bontrager reminded the public not to worry, although the AI ​​has the ability to crack artificial fingerprints. According to Bontrager, this method is still difficult to accept and needs further development to be used in the future.

“This success rate applies only to fingerprint scanners on mobile phones. It takes a lot of effort to try to create such a system, ”he said.

According to ZDNetOn Friday (11/16) DeepMasterPrints improved with MasterPrint, which was developed by researchers at the School of Engineering at New York University and the College of Engineering at Michigan State University. MasterPrint explores the security vulnerabilities of the partial fingerprint matching system.

At that time, the technology depended on changing the details of the fingerprint image that was taken and used by the fingerprint scanner for the relevant purpose.

The previous method was able to simulate images stored in files (file), but can not create realistic images of fingerprints, such as the original without stains.

The focus of the study is to prove that many fingerprint scanning devices read only part of the surface of a fingerprint. Although some different parts of the tip of the finger have more in common than others.

Although research aims to encourage companies to improve fingerprint protection efforts. Without verifying that fingerprints scanned with biometrics come from real people, the threat of burglary using fingerprints can become a reality.


Source link