Peace sign selfies may be revealing more than just a casual pose. Experts say AI-enhanced images can, in some cases, help reconstruct fingerprint data from photos.
As someone who used to live in Taiwan, I noticed a strong tendency for the citizens to use peace signs when taking selfies. However, I never actually considered that my security could be at risk, as I often replicated the gesture to try to culturally assimilate.
And recently, Chinese financial expert Li Chang demonstrated on reality TV that a high-resolution camera can capture and extract fingerprints, especially within 1.5 meters.
On the show, Chang showed that even when the fingerprints seem blurry to the naked eye, AI software can be used to sharpen the image in order to obtain the data.
But how likely is this to actually happen? “I do think it’s technically possible to reconstruct fingerprints from photos now, but it’s honestly a pretty targeted threat,” said Lewis Berry, a principal security architect and Microsoft MVP at Inforcer.
The issue has been discussed for over a decade, as in 2014, biometrics researcher Jan Krisller demonstrated how easy it was for him to access the fingerprints of then German Defense Minister Ursula von der Leyen while she was speaking live at an event.
Targeted scenarios
Fingerprints are not only used to access smartphones, but also on a large scale elsewhere. Think warehouses, even data centers, and border control – the implications could be catastrophic.
And while we may grimace at the thought of a blurred selfie being used to crack into any of the mentioned locations, it may be individuals higher up the ladder that ought to be more concerned:
“This becomes a lot more relevant in targeted scenarios like higher profile individuals, perhaps repeated exposure of the same finger," explained Berry.
And in his briefings online, Chang advised us to blur our fingerprints when taking photos and to avoid repeating the peace gesture, especially since hackers often need a few examples to work with.
Your email address will not be published. Required fields are markedmarked