A great talent for evil and a handful of snapshots were enough for the Israeli scammer Simon Leviev to get millions of dollars through Tinder. Apparently, his life was a mansion overlooking the sea: he flew in jets privates, drove high-end cars and rested in fabulous hotels. Shared on the dating app, the pictures of him looked real. However, spiced up with a little love they ruined dozens of women They did not hesitate to lend him money that he never returned. The documentary film tells The Tinder scammerpremiered on Netflix, and repeats it in his teaching talks Fernando Pérez-González, professor at the University of Vigo who heads a research team that has given birth to a software called Fawrensian. The program is capable of detecting the manipulation of photographic files to prevent cyberfraud and can verify more than a million documents per day. It is a technology transformed into a forensic analysis kit of multimedia content whose license is used by companies (some from Silicon Valley) in numerous business processes, for example, in which personal documents are involved.
It all started 15 years ago with a question. “We came from working on image watermarking, which basically consists of hiding information in photographs to protect the copyright and things like that”, explains Pérez-González in his office at the university. They then wondered how they could find out if a photograph had been taken by a certain camera. “We’re talking about the specific camera, not just a model.” They discovered that the sensor of each camera has imperfections that are reflected in the images, although no one can detect them with the naked eye. “It’s kind of like a fingerprint.” That fingerprint can be extracted from a group of photos taken by a camera, including those from mobile phones. “From there you can resolve the question of whether a given image has been taken by one camera or another.”
This has expert applications. For example, if the police seize a pedophile’s hard drive, they can determine how many different cameras have taken the images it contains. It is enough to have enough snapshots, because not all of them are equally good at extracting a print. This process translates into probabilities, brackets similar to those offered by DNA tests to determine if a genetic sample belongs to an individual. “It’s pretty amazing. There are those who call it camera ballistics, because it is very similar to the analyzes that say if a gun is the author of a shot.
But forensic tools gave way to others more demanded by the market. “We started thinking about uncovering fake newsbut where there really is money is in the processes called Know your customer (KYC), where, for example, you ask for a loan and they ask for an image of your card. There is a lot at stake there and companies need new tools.” Its technology looks for alterations in the properties of the documents to detect inconsistencies. Many examples circulate on Twitter, such as a photo of Pope Benedict XVI kissing an imam on the mouth for a Benetton campaign that was withdrawn. “When you do a jpg compression, for example, a series of very specific properties are induced in the image. We detect traces of that double compression. It is true that many social networks recompress the photos we upload, and there is nothing wrong with that. But if you send a photo of your ID to a bank, there shouldn’t be a double compression. We warned you.”
A filter for Twitter?
David Vázquez, a researcher who is part of the team, takes five seconds to change a number on his ID on the screen. The trap is not noticeable. Filtered by the program, the image is shortly after converted into a heat map that shows how likely each pixel is to be manipulated, and his deception is uncovered. He takes other images from Twitter, processes them, and again detects manipulation in one posted about the protests in Catalonia in 2019. A priori, it seems that many people would pay to use this software on-line to check photos from Tinder or Twitter, but the researchers do not market it to individual users. “It would be much more complicated for us to manage. How many times would each customer use it? Surely they would not be willing to pay. We would have to have a customer service, it would be another business model different from the one we are following”, they say in the university team.
Fawrensian, developed with the support of the Xunta, is only sold as a license to other companies and five people from the Center for Research in Telecommunications Technologies of the University of Vigo work on it (Atlantic). They hope that the revenue from the product can be returned to its public sponsors: “The goal is that this is used and that we can continue to solve other problems. And that the administration continues betting on taking technologies out of the university and putting them on the market”. In a few years they believe that they will recover the 350,000 euros that it has cost.
Exclusive content for subscribers
read without limits