Possibility of Fingerprint Reconstruction from Photos Using AI Technology
Chinese Expert: "Higher Resolution Photos Increase the Risk"

There has been a warning that the commonly used 'V' pose when taking photos could lead to the risk of personal information leakage. With the advancement of artificial intelligence (AI) technology and high-resolution cameras, it has been pointed out that fingerprints in photos can potentially be reconstructed using only the fingers.


According to the South China Morning Post (SCMP) in Hong Kong, financial expert Li Chang appeared on a Chinese television program and demonstrated the possibility of reconstructing fingerprints by using celebrity selfie photos.


Image extracted by AI of a person taking a photo with a 'V' pose

Image extracted by AI of a person taking a photo with a 'V' pose

View original image

Li Chang explained that if a photo is taken from within 1.5 meters with the fingers facing directly toward the camera, it is possible to extract fingerprints. He added that even at a distance of 1.5 to 3 meters, about half of the hand’s detailed features could be restored. The program showed a process where, after retouching the photo using editing software and AI tools, the ridges of the fingerprints became visible.


Jing Jiu, a professor of cryptography at the University of Chinese Academy of Sciences, told China Newsweek that even a 'scissors-hand' pose in a high-resolution portrait could allow for the reconstruction of detailed hand information.


However, Professor Jing Jiu explained that due to various variables such as lighting, focus, and photo clarity, actual fingerprint reconstruction is not easy. Nevertheless, he pointed out that the risk could increase if high-resolution devices are used more frequently and multiple related photos are leaked.


Li Chang warned that fingerprints, like facial information, are permanent biometric data and are difficult to change once leaked. He explained that such leaks could lead to financial losses or identity theft. He further recommended blurring or pixelating the hand area before posting selfies online, and advised against registering fingerprint information on devices of unclear origin.


After the program aired, the topic spread rapidly on Chinese social networking services (SNS). One user commented, "From now on, I guess we should just make a fist in selfies." Another user wrote, "I should wipe off fingerprint marks from my phone screen more often."


According to Chinese media Sohu News, there was an actual incident in July last year in Hangzhou, China, where a man posted a photo online with his fingers clearly visible. A criminal organization attempted to download the photo and use it to unlock a smart door lock. However, the attempt was ultimately unsuccessful.


Meanwhile, Fei Zeyong, director of the Qianxin Industrial Security Research Center in China, stated that even if it is technically possible to extract fingerprints from selfie photos, it is not easy to use them in actual crimes. Fei explained that fingerprint information is primarily used in environments with relatively low security levels and is more likely to be exploited in the creation of forged documents, among other things.


Fei also explained that AI technology is being used to counter new types of fraud, and that some AI tools can detect face synthesis in real time.


Meanwhile, as AI-based image processing technology becomes more advanced, personal information leakage through photos is emerging as a new issue. On Chinese SNS, concerns have been raised that apps such as 'AI Physiognomy Test' or 'Free Palm Reading' could collect, store, or misuse facial and palm biometric information.



In fact, an employee at an IT company in Shanghai was caught after illegally collecting and storing more than 1,700 pieces of facial biometric information through a 'beauty test' application and selling them on the dark web.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily(www.asiae.co.kr). All rights reserved.

Today’s Briefing