The social media behemoth, Facebook is finally letting the world take a peek behind their closed doors when it comes to how they are operating. Recently, the company has opened up about how they are combatting the creation of fake accounts on their social media platform, using internal tools.
Deep Entity Classification (DEC) is an artificial intelligence-powered tool that is being used by the company at present. It has been reportedly been an extremely instrumental and effective tool to battle and terminate fake accounts on Facebook.
The company has claimed that the Deep Entity Classification, aka DEC, has helped them reduce 27% of the estimated volume of accounts related to spam and scam.
Besides being able to spot and take into account the activity of a suspect account, the Deep Entity Classification or the DEC also is capable of evaluating all the surrounding information. This includes the behaviour of the kind of interactions the suspected accounts and pages have with other accounts.
A whopping 6.5 billion fake accounts that were either tried to be created or created successfully last year by various scammers and malicious agents are taken down by Facebook with the help of the DEC.
With the help of this artificial intelligence-powered tool, Facebook’s automated systems manage to discover these accounts during the signup process itself. Even for the ones which tend to slip past the creation process get quickly gets found and terminated before being reported by a real user.
According to the social media giant, they estimate that among the over 2.5 billion monthly users who are currently active on the platform, 5% is comprised of fake accounts.
Fake accounts are typically created in order to facilitate various malicious scammers, spammers, and people attempting to phish vulnerable users get their hands on the personal data of other users on the platform. That data is then further used in various criminal activities of financial or identity theft related nature.
This is where Facebook’s automated system equipped with the DEC comes in to solve this crucial problem. This tool takes into account over 20,000 features per profile to analyze user behaviour. Therefore, the DEC ensures a sophisticated and holistic approach is being taken.
The tool doesn’t only take into account the suspicious user profile itself but also the activity of an account which the suspicious account has sent a friend request to or is already connected to. This helps the tool detect how these malicious accounts repeatedly try to mimic genuine behaviour on the platform.
However, it is believed by Facebook that spammers will start taking note of this and they will find a way around it or try and get better at mimicking to be genuine and real users on their platform.
Facebook Fake Accounts Issue in 2020
Last year, AI tools were reportedly used to generate genuine-looking profile photos. These were then used on Facebook and Twitter to create fake accounts which pushed massive pro-Trump messaging. Of course, both the social media platforms managed to shut most of them down but it wasn’t an easy task for them to do. As these photos weren’t scraped from the internet but generated using neural networks, it was hard to flag these accounts as fake.
This time Facebook is trying to stay one step ahead by using this tool before the upcoming 2020 US Presidential elections. Besides a much needed precautionary measure, this move can also be observed as a strategic way to start gaining the confidence of their user base and the US government after their huge lapses in privacy and content regulations that were constantly at the centre of the spotlight throughout the last two last years.
However, even though it is a good starting point for taking baby steps, they still have a long way to go as their tussle with privacy is still on in other areas such as with their newly launched dating app.