Facebook has revealed a security system called FIS (Facebook Immunity System). FIS controls every read-write action of each user on the social network. This defends 99% of Facebookers from spam, as stated by Facebook itself. The defense system happens to be the biggest in the world, equipped with 30 supervisors, enabled to learn and take actions on noticing suspicious behavior patterns.

The system, as it seems to me, is still far from the artificial intelligence. But is already able to combat massive viral script attacks within the network by blocking messages based on keywords once found and identified in spam chat messages.

Still unknown techniques make the system vulnerable. Social bots tend to fool the system, even adding thousands of friends and harvesting sensitive data like e-mail addresses. Due to peaking number of friend rejections spam bots receive, Facebook predicts improving FIS based on higher, than usual rejections amount.

I personally recently found out Facebook controls the speed of friends addition, if you add them too fast - you get considered to be a bot, even if you just re-add known people to the newly created account. If some people reject you, you already get blocked for a couple days, or a week without even a guess, who did it.

Still much job to do, but unless real artificial intelligence gets invented, we are going to face constant rules creating and breaking in this highly socialized world.