Beware of who you befriend on Facebook as it might be a bot

Nov 3, 2011 12:27 GMT  ·  By

Researchers proved that a small army of socialbots can easily infiltrate Facebook by fooling all the protection mechanisms into thinking they are actually human users.

In a paper called “The Socialbot Network: When Bots Socialize for Fame and Money”, scientists revealed how they created 102 bots that were designed to be seen as humans by social networks. By using some cleaver scripts they managed to almost perfectly replicate any actions that would be normally performed by regular customers.

They chose Facebook to act as the OSN (online social network) but they believe any social media website is susceptible to a similar approach.

The eight-week experiment showed that the socialbots represented on Facebook by a name and a picture managed to generate an incoming traffic of 250 gigabytes and an outgoing traffic of 3 gigabytes.

With the help of a botmaster, the socialbots and a command and control center, they were able to harvest user data from more than 3000 profiles from the direct neighborhood and 1 million profiles if we consider the extended neighborhood of the human-like bots.

To make sure they don't trigger any alarms, the bots were programed to send only 25 friend requests each day, two weeks since the beginning of the experiment close to 1000 people accepting the invitations.

The experiment shows that profiles can be created by using an automated mechanism, the CAPTCHAs used by Facebook in many cases not representing an impediment. Their Immune System was also no match for the socialbots, only 20% of them being blocked.

“We have evaluated how vulnerable OSNs are to a large-scale in ltration by a Socialbot Network (SbN). We used Facebook as a representative OSN, and found that using bots that mimic real OSN users is e ective in in ltrating Facebook on a large scale, especially when the users and the bots share mutual connections,” concludes the study.