Symantec: Google Groups Used to Send Commands to Malware

Symantec researchers have uncovered a Trojan using a private newsgroup within Google Groups as a command and control server. The move follows an attempt to use Twitter as a C&C earlier this year.

Symantec has uncovered a scheme to use a Google Groups newsgroup to sneak commands to malware on compromised computers.

The move is another example of attackers looking for covert ways to communicate to their bots. Earlier this year, attackers were found using Twitter as a command and control (C&C) mechanism. By integrating their messages with legitimate communications, attackers make it more difficult to identify and shut down their C&C, according to Symantec.

"This technique is analogous to the use of encoding messages in newspaper ads that were commonplace in spy novels," Zulfikar Ramzan, technical director of Symantec Security Response, told eWEEK. "What attackers are taking advantage of are online mediums that allow pretty much anyone to post content and are both highly available as well as readily accessible from the outside. I believe they are going down this route, since it represents a very easy and inexpensive avenue for setting up command and control."

In this case, the attackers infect users with a Trojan Symantec calls Trojan.Grups. The malware targets Windows systems and installs a backdoor on infected machines. When it's run, it logs into a private newsgroup on Google Groups and reads a message on the newsgroup that contains the commands to be carried out. Afterward it posts a response to the newsgroup stating whether the command was successfully executed or not.

"One noteworthy aspect of this attack is the use of the RC4 stream cipher to encrypt the messages being passed back and forth," Ramzan explained. "While encrypting communication is a conceptually simple thing to do, it demonstrates that attackers are trying to take extra measures to avoid detection and also to potentially avoid having their botnet overtaken by some other rogue party."

The method has some drawbacks for the attacker, however, as every response is stored as a posting in the newsgroup, making it possible to backtrack the Trojan's activity in detail. Symantec researcher Gavin O. Gorman speculated that given the relatively low amount of activity by the Trojan-some 3,000 newsgroup posts since November 2008-and an examination of its code, this may be a prototype implementation to test the Web-based newsgroups as C&Cs.

"It is most likely Taiwanese-based since the newsgroup language is Chinese [simplified], with several references to .tw domains in commands," Gorman blogged. "The low numbers imply this is a discreet Trojan, used to subtly gather information and potentially determine future attack targets. In addition, there is no attempt within the DLL to maintain persistence on the attacked computer, further evidence of a Trojan attempting to remain undiscovered. Such a Trojan could potentially have been developed for targeted corporate espionage where anonymity and discretion are priorities."

By going this route, attackers don't have to incur the costs of setting up an explicit command and control server, Ramzan added.

"I expect that we'll continue to see these types of attacks, and that attackers will develop more refined approaches as sites like Google and Twitter develop better detection and containment mechanisms," he said.