In a YouTube video he has published on his blog, Zilberman shows how a hacker (dubbed "Joe Evil") can steal information from a Facebook user (who Zilberman calls "Cookie Monster") by planting an IMG tag on a Website message forum.
Worrisome, the attack works even if the Facebook user has applied strict privacy settings to protect his data. This is because Zilberman found a way to trick Facebook into thinking that the Facebook application page his IMG tag on a third-party Website is accessing is a result of a logged-in Facebook user's interaction.
In effect this is a cross-site request forgery (CSRF) attack, fooling Facebook into thinking that the user is sending it commands rather than a third-party.
Remember, the only thing that the innocent Facebook user has done is visit a legitimate third-party Web forum where an image is displayed.
The way in which a hacker achieves this attack is beyond the scope of this blog entry, but you can read more in another of Zilberman's blog entries.
Ronen Zilberman says the specific vulnerability he demonstrates in the video has now been patched by Facebook, but warns, "It's likely that it is still possible to launch this type of attack using other mechanisms and other social networks."
It really is time for the big social networking sites to examine their systems and determine how, now they have gathered a huge number of members, they are going to protect us all from virus writers, identity thieves, spammers, and scammers.
The honeymoon period for these sites is well and truly over. Our personally identifiable information is at risk as a result of constant attacks that the Web 2.0 sites are simply not mature enough to protect against.
Graham Cluley is senior technology consultant at Sophos, and has been working in the computer security field since the early 1990s. When he's not updating his other blog on the Sophos website, you can find him on Twitter at @gcluley. Special to Dark Reading.