Security researchers have spotted a surge in attacks against online banking customers, thanks to a new strain of Java-exploiting Trojan Caphaw (aka Shylock).
Over the last month or so the malware has targeted customers in at least 24 financial institutions, including Bank of Scotland, Barclays Bank, First Direct, Santander Direkt Bank AG and Capital One Financial Corporation, according to security researchers at cloud security firm Zscaler. There's no word on whether or how successful its attacks have been and which bank's customers have been affected. Caphaw (Shylock) is most active in the UK, Italy, Denmark and Turkey.
"We have detected hundreds of infections, but there is no way to calculate the losses," Zscaler researcher Chris Mannon told El Reg.
The Trojan hooks itself into the browser processes of victims before using a self-signed SSL certificate to trigger encrypted “phone home” communication with remote command and control servers. This encryption is designed to keep the malware under the radar of corporate and ISP-level network security tools. Detection by endpoint security scanners is also low, according to Zscaler.
Caphaw appears to be spreading using a Java exploit from compromised websites as part of a drive-by download attack. However evidence for this theory remains circumstantial, as an advisory from Zscaler explains.
"At the time of research, we were unable to identify the initial infection vector," Mannon and fellow Zscaler researchers Sachin Deodhar explain in a blog post.
"We can tell that it is more than likely arriving as part of an exploit kit honing in on vulnerable versions of Java. The reason we suspect this is that the User-Agent for every single transaction that has come through our Behavioral Analysis (BA) solution has been: Mozilla/4.0 (Windows XP 5.1) Java/1.6.0_07."
Mannon added: "We suspect it is coming from a Java exploit on the version listed in the blog. Other vectors this threat has used in the past include Skype, social media, and email spam."
Caphaw features a domain generation algorithm that generates a large number of quasi-random domain names that are then used to "dial home" and receive/send commands/data. This is far from a new tactic in botnet administration but it's still a successful approach in making life difficult for law enforcement.
"The large number of potential rendezvous points with randomised names makes it extremely difficult for investigators and law enforcement agencies to identify and 'take down' the CnC [command and control] infrastructure,” said Mannon. “Furthermore, by using encryption, it adds another layer of difficulty to the process of identifying and targeting the command and control assets." ®