Eliran Shachar was browsing an online dating site, when one woman piqued his interest. They got to chatting, and Shachar decided to check out her Facebook profile to find out some more information.
After a week of this budding relationship, Shachar realised that there was something fishy about this woman’s Facebook profile and decided to investigate. He discovered that she had 20 different profiles with the same photos, and when he confronted the woman, she proceeded to delete all of the accounts in three seconds.
Shaken by the experience, Shachar realised that this phenomenon of fake profiles on Facebook was more common than he previously thought. And it wasn’t just potential matches on dating sites, there were also criminals leveraging fake profiles to connect with victims. While Facebook’s annual report claims that only about 0.4% – 1.2% of all active users are abusive accounts that create fake likes, Shachar claims that about 20-25% of Facebook accounts are fake.
As a web developer with more than a decade of startup experience, Shachar decided to do something about this. And so FakeOff came into existence. FakeOff is an app that syncs to your Facebook account and can alert you to users in your friend list that are suspicious.
This is potentially huge since fake accounts have become a big problem on Facebook. They can negatively impact companies when their advertising spend is being wasted on fake accounts that have liked their page. Raaj Kapur Brar, the CEO of fashion web site publisher Fetopolis, claims that fake accounts destroyed his advertising business on Facebook, forcing him to end his relationship with the social media platform.
FakeOff attacks these kinds of problems head on with an algorithm that detects suspicious accounts and alerts Facebook users of fraudulous accounts.
A free version of the app will look at 10 days of timeline information on a user, and the full version goes back a whole year. After going through the data, FakeOff gives the user a rating of 1-10, 1 being most likely a fake account and 10 being most likely a real account.
“Our app is not only checking photos, but we also built an algorithm to check the timeline so we can know what is normal and what isn’t normal when someone speaks,” Shachar told Business Insider. “We can monitor the complexity of the conversation and to see if it’s logical or if there’s something strange there. We collect a lot of information on the timeline. We see if he’s using Instagram. We collect a lot of information and give him a score.”
Another sign that hints at a fake profile is when a profile frequently tries to engage other users in private messaging as opposed to public posts. When a profile has synced its posts with a phone or Instagram account, this signals that it is most likely a real profile since it would be difficult to sync multiple different fake accounts.
As it goes through different searches, FakeOff is putting together a larger database of verified and fake profiles to aid its algorithm for future searches. Through a pilot in India, the app acquired 80,000 users but as it grows, it hopes to gather more data and help spot fraud more effectively.
While FakeOff is currently focused on detecting fake profiles, Shachar said that he is developing a similar algorithm to detect fake likes, a product that would most likely be geared towards the professional or corporate world. This would be able to point to the amount of fake — or bought — likes on Facebook pages. For instance, during the pilot in India, it happened to be election year and one of the candidates had about 2 million likes on Facebook. Shachar used this algorithm and estimates that 30% of those likes were not real.
FakeOff also may look to expand to other social media platforms, but since Facebook has such a massive amount of users, Shachar figured it was a good place to start.
For now though, FakeOff is focusing on protecting individual users.
“The main goal for now for this app is for singles that usually add new friends every week and of course teens because teens are vulnerable to sex offenders and this is very big problem for teens because some of them are a little bit naive,” Shachar said. “We really look at it like an antivirus, the same way you want to protect your computer from viruses you want to protect your profile from fakes.”