A team of researchers have built a malicious Facebook program an experiment to demonstrate the possible dangers of social networking applications.
The experiment shows the ease with which attackers could dupe large number of users into downloading a seemingly harmless application that actually performs a clandestine attack that can cripple a Web site.
Facebook and other Web sites such as MySpace, Bebo and Google are creating technology platforms that let third-party developers build applications to run on those sites. The concept has opened the door to innovation, but also prompted worries over how those applications could be used for spam or steal personal data.
The researchers developed an application called “Photo of the Day,” which serves up a new National Geographic photo daily. But in the background, every time the application is clicked, it sends a 600 K-byte HTTP request for images to a victim’s Web site.
Those requests, as well as those images, are not seen by someone using Photo of the Day, which the researchers have termed a “Facebot” application. The effect is a flood of traffic to the victim’s Web site, known as a denial-of-service attack.
The researchers uploaded their application to Facebook in January and told a few colleagues about it. Even without advertising or other promotion, close to 1,000 people installed it in their profiles, much to the researchers’ surprise.
They then monitored traffic on a Web site they set up for Photo of the Day to attack. If those traffic figures were applied to Facebook applications that have a million or more users, they estimated a victim’s Web site could be bombarded by as much as 23 M bits per second of traffic, or 248 G bytes of unwanted data per day.
“Facebook applications have a highly-distributed platform with significant attack firepower under their control,” wrote the researchers.
The malicious Facebot could also be rigged for other nefarious duties. An attacker could create an application that uses JavaScript and HTTP requests to figure out if a particular host has certain ports open, they wrote. Another possibility is to construct an application that delivers a malicious link in order to infect a Web site with malware.
Since Facebook applications can get access to users’ personal details, it would also be possible for the application to grab all of those details and post them to a remote server, they wrote.
However, social networking sites can take measures to prevent bad applications, the researchers said. One remedy is ensuring that applications can’t interact with hosts that aren’t part of the social network. New applications should also be vigorously verified by the social networking site. APIs (application programming interfaces) should be crafted so as not to allow too much interaction with the rest of the Internet.
Photo of the Day is still listed on Facebook, with its authorship attributed to Andreas Makridakis, one of the researchers. The application has 543 users now, with several comments praising it.
The study was published by the Foundation for Research and Technology in Heraklion, Greece, and the Institute for Infocomm Research in Singapore.