Memes on Twitter Used to Communicate With Malware

A new and otherwise ordinary malware tool is garnering some attention from security researchers for its ability to retrieve malicious commands via code hidden in a couple of Twitter memes.

The malware (TROJAN.MSIL.BERBOMTHUM.AA) targets Windows systems and, like more than 90% of all malicious code, is distributed via phishing attacks. Once installed on a system, the malware can perform several common functions like capturing local screen shots, enumerating applications on the system, checking for vulnerabilities in them, capturing clipboard content, and sending files back to the attacker.

What's noteworthy about the new Trojan is its use of the Twitter memes to retrieve malicious instructions, according to Trend Micro, the first to report on the threat. The authors of the malware—currently unknown—posted two tweets featuring the malicious memes in late October using a Twitter account that appears to have been created last year.

Embedded in the memes is a /print command that basically instructs the infected computer to take screen shots and perform other malicious functions. The malware extracts the command after first downloading the malicious memes to the infected system. The malware supports a variety of other commands including /processos for retrieving a list of running processes, /clip for capturing clipboard contents, and /username for grabbing the username from the infected system.

The screenshots and other captured data are then sent to a control server whose address the malware obtains via a hard-coded URL on pastebin.com, Trend Micro said in a report on the attack.

This is not the first time that attackers have used steganography—a method for hiding code and malicious payloads in images—to try and sneak malware and malicious activity past threat detection tools. Neither is the new Trojan the first example of malware using a popular social media platform for command and communication purposes. In fact, Trend Micro itself last year had warned about the growing abuse, by cybercriminals, of chat platform APIs for command and communication purposes.

But by combining the two techniques—steganography and the use of Twitter—the attackers are making it harder for defenders to discover and take down the malicious activity says Mark Nunnikhoven, Trend Micro’s vice president of cloud research.

TROJAN.MSIL.BERBOMTHUM.AA is a relatively standard piece of malware. "There's nothing particularly unique about it beyond how it is set up to retrieve commands," Nunnikhoven says. "If the infected organization is analyzing network traffic, it's unlikely that they would count traffic to and from Twitter.com as out of the ordinary."  

Hiding in Plain Sight

Using the steganography technique, the attackers are hiding their C&C channel in a stream of legitimate activity associated with a commonly used social media tool, he notes. "The takeaway for enterprises is that attackers are growing more sophisticated with how they avoid traditional security techniques and controls," he says.

Trend Micro has no specific intel on the attacker and its motivation, nor are there hints or indicators in the malware itself that make attribution possible, Nunnikhoven adds.

Travis Smith, principal security researcher at Tripwire, says the use of social media for command and communication is not uncommon. However, sophisticated attackers tend to avoid it since it's a known technique and relatively easily detectable via social media scanning tools. At the same time, malicious code hidden in images posted to social media can be hard to detect, he says.

"When you do not own the original image that is being used to hide data within, it can become increasingly difficult to detect the usage of steganography," especially when the payload is small, such as the test for C2, he says.

Currently there are no tools available that continuously scan social media images for malware. "The amount of images being posted to social media on a daily basis would make that type of analysis impractical," he says.

Related Content: