Fakebook’s evolving AI and machine learning model took down 6.6 billion fake accounts in 2019. That’s no mean feat when it claims it has 2.5 billion legitimate ones.
The tool is called Deep Entity Classification. It analyses active Fakebook accounts and the individual’s behaviour in the Facebook community. That includes things like the number of friends requests an account has sent and the groups or pages it has joined or followed.
Importantly the machine learning model retrains itself automatically, so it can evolve as spammers adapt. And they do. Fool Facebook is an ever-evolving game.
Facebook data science manager Bochra Gharbaoui said, “Adversaries move fast. Their adaptation cycle is fierce, and it’s getting more sophisticated.”
Gharbaoui estimates that about 5% (125 million) of its 2.5 billion active accounts are fake.
Analysts suggest that the real number could be 100% higher if you count cloned and duplicate accounts.
The massive number of accounts are largely driven by automated attacks by bad actors who attempt to create large volumes of accounts at one time.
Fakebook accounts are a scourge
We have been reporting on Facebook f’ups since 2005! Hundreds of articles all warning you that the Facebook drug is addictive and highly dangerous. We are now card-carrying #Delete Facebook members.
Let’s not even go to the ‘conspiracy’ claims that over half of Facebook’s active accounts are fake anyway – mere rubbery numbers to keep advertisers on the platform.
The admission by Mark Zuckerberg last month that it removes over one million fake accounts every day is an inditement on Facebook, its heritage and its University programming pedigree. It is also an inditement on Zuc’s “Learn fast and fix things” philosophy that has led to the Cambridge Analytica scandal and so much more. He is so good at stuffing up and asking forgiveness later.
In short, if Facebook had been developed by a responsible entity and subject to practical privacy legislation, this would not have happened. Unfortunately, you can’t change history.