How to Spot a Deepfake

A senior vulnerability risk manager at Blue Cross Blue Shield, Wayne Nordstrom researches cybersecurity trends and emerging technology. Among others, Wayne Nordstrom focuses on coming up with solutions to fight cyber fraud.

One of the latest methods by which scammers and hackers commit cyber fraud is using the technology deepfaking, which can digitally manipulate a person’s appearance and identity. The most famous deepfakes belong to popular personalities, morphing their faces to someone else’s as they make an impression. A more sinister use of the technology has also emerged, with scammers using the algorithmic technology to produce more subtle deepfakes to manipulate victims.

These deepfake scams can take a variety of forms, from blackmail to extortion schemes. However, researchers have found a few ways to spot a deepfake, even with the naked eye. The tell-tale sign, studies show, is the pupils of a deepfaked image. While other features can be approximated by the algorithm, even high-quality deepfakes have trouble maintaining a clear, round shape for a person’s pupils, leading to said pupils having a distorted or ragged edge.

There are some issues with this detection method, not least that some real people have irregular pupils due to genetic factors but also that all the research on the signs of deepfakes is in the public domain. Like the readily available deepfake technology, the manual on how to spot them is in the hands of those who would use it to manipulate others. This has led some researchers to worry that deepfake algorithms could be improved by referencing research on their weaknesses, thus requiring new detection techniques to be discovered.

from WordPress https://ift.tt/3mzezLs
via IFTTT