The Chrome Extension can identify fake images on user profiles with an accuracy of up to 99.29%.
According to Petapixel, V7 Labs has developed new software based on artificial intelligence (AI) that functions as a Google Chrome extension. It is capable of detecting profile pictures created artificially—similar to the images you see above—with a claimed accuracy of 99.28%.
This extension can detect profile pictures generated artificially.
V7 Labs is a software company specializing in designing products to automate visual tasks. The company states that people currently rely heavily on images to make decisions (for example, becoming friends with a stranger on social media after seeing their profile picture). They are working towards making deep learning technology more powerful, allowing any business to implement modern AI technology to recognize images from a single platform.
Detecting Fake Images on Profiles
Alberto Rizzoli, one of the founders of V7 Labs, describes the new software as designed to help limit misinformation in the online environment. He demonstrated this feature in a video on the Loom platform.
Rizzoli stated: “Today, there is a lot of misleading content on social media, and many people share this content using fake profiles. We have created a Chrome extension that can detect whether a profile picture is a real person or an AI-generated portrait.”
He also showcased how the extension works by checking the profile picture of someone who recently added him on LinkedIn. With just one click, the extension quickly identified that she was a fake. In that specific example, there were several signs indicating the photo was not real, such as a messy earring and oddly looking pupils. However, these details were only noticeable upon close inspection, and the fake profile still fooled many internet users.
Rizzoli continued with a second example where the extension’s identification accuracy was nearly 100%.
He explained that many of these images were created on the website This Person Doesn’t Exist. This site uses an AI face generator powered by NVIDIA StyleGAN, a neural network developed in 2018. The website was created by a software engineer from San Francisco and is very simple to use. When the page loads, it displays a new face generated by AI. Each time the page is refreshed, a new face appears (these faces are created by AI from a 512-dimensional vector).
Combating Dangerous Misinformation
Some of Rizzoli’s initial examples may seem relatively harmless, but he claims that his software can better combat fake news and misinformation.
He noted: “This is something that is developing quite a lot, especially on Amazon, where many authors sell books. Malicious actors can easily create a fake profile and market counterfeit books at lower prices than genuine ones.”
Of course, the issue of impersonating others also occurs on social media. Rizzoli pointed to an example of a Twitter user with an AI-generated face spreading misinformation about the war in Ukraine. This information has a very ‘negative’ impact on economic and social life.
He told Petapixel: “This Chrome extension will help authorities (and ordinary people) detect and report profiles spreading fake news. These profiles create chaos in information, especially regarding issues like the conflict between Russia and Ukraine.”
Currently, this extension only works on images generated by GAN. Therefore, it cannot detect high-quality images in videos (at least not at this time).
