Social networks fight against all odds to prevent the registration of false accounts in their respective services, making use of their own technologies that act as gatekeepers, capable of detecting and eliminating false accounts in a short period of time.
But these technologies are not infallible and there are always registrations and uses of false accounts that are sneaked as legitimate accounts, being increasingly complex to identify them due to their surprising realism with respect to the accounts of real people.
It is increasingly complex to detect fake accounts
Fortunately, the human factor also helps identify fake accounts that have escaped the increasingly complex control process that tries to prevent them. And it is that, among other things, false accounts usually make use of faces generated by Artificial Intelligence , so it will be the behaviors that can lead to suspicion.
This is what is reflected in a recent transparency report published by LinkedIn:
Our automated defenses blocked 97.1% of all fake accounts we stopped during the period from January to June 2021. We continue to invest in manual and automated defenses to prevent or remove malicious LinkedIn accounts.
Said report specifies of the 15 million accounts detected in the first half of 2021, most of them were detected during the registration process, although another important part of them were detected and eliminated shortly after through automated systems.
Despite this, there have continued to be tens of thousands of false accounts that had to be removed manually thanks to the multiple reports received from users based on the suspicious behavior of said accounts.
Renée DiResta and Josh Goldstein, researchers at the Stanford Internet Observatory, were part of those users who have sent different reports about suspicious user accounts, having found more than 1,000 fake accounts within the social network itself as part of their investigations, assuming a part of the 90,000 false accounts that were found and finally eliminated thanks to this procedure.
Although the number of false accounts that have managed to evade the control system may seem low, the trend found by those responsible for LinkedIn is that it tends to increase, being increasingly complex to detect and eliminate them by automatic systems, which also they have come to advance over time, so together with this work, they also continue to fight to improve detection and manual removal.
The most curious thing is that companies specialized in generating realistic avatars using Artificial Intelligence have also been detected. Two of these companies were also present and have been removed from LinkedIn.