Facebook experiment helps battle man-in-the-middle attacks

Facebook grad
Credit: MKH Marketing

Facebook, Carnegie Mellon University detection tool finds tampered or forged certificates in more than 6,800 SSL connections to the social network

Researchers at Facebook and Carnegie Mellon University have developed a detection tool for man-in-the middle attacks that security pros might find helpful in protecting corporate data.

While the technique has been around for years, the researchers were able to adapt it for use on Facebook, proving that the method would work on a large-scale network.

[DDoS attack from Facebook possible, but unlikely]

The researchers analyzed more than 3 million SSL connections to the website and found 0.2 percent, or 6,845, contained tampered or forged certificates.

Most of the changes were related to anti-virus software and corporate content filters, with only 121 forged by malware and 330 by adware, the researchers said Tuesday.

"The takeaway for IT professionals is that SSL man-in-the-middle is at the same time more common than expected, but also mostly perpetrated by the good guys," Wolfgang Kandek, chief technology officer for risk management company Qualys, said.

A man-in-the-middle attack is a form of eavesdropping on communications between a browser and a web server. If the traffic is encrypted through the widely used SSL protocol, then the attackers will somehow tinker with the certificates to decrypt the communications.

To detect certificate shenanigans, the researchers embedded a Flash applet in web pages served to Facebook users chosen at random. The code bypassed the network protocol stack of the browser and sent information on the certificates to a server run by the researchers.

In proving the technique works on a large scale, the experiment, conducted under Facebook's data use policy, showed that the method would be useful to corporate security pros who wanted to watch for man-in-the-middle attacks on users of company websites, Collin Jackson, co-author of the study and an assistant research professor at Carnegie Mellon, said.

"We were looking at Internet users as a whole, but CSOs are probably a lot more interested in protecting the people inside their organization," Jackson said. "This would be one way to identify if any employee's traffic is being tampered with, at least when they're communicating internally,"

The research also highlighted the potential security risks introduced by anti-virus products and content filtering technology that act as proxies through which all Internet traffic flows. These products can be on a person's computer or in the cloud.

Either way, the software uses its certificates in place of the browser's, which could weaken the latter's ability to validate certificates, making communications with web servers less secure, Jackson said.

[FTC warning unlikely to stop Facebook from changing WhatsApp privacy policies]

Bottom line, companies using such proxies have to trust their ability to manage certificates in an environment that's more complex than if the vendors somehow leveraged the browsers' certificates, Jackson said.

"It's really better for privacy and security to let the browser take care of that," he said.

Join the discussion
Be the first to comment on this article. Our Commenting Policies