This article is more than 1 year old
Twitter transformed into botnet command channel
Victim becomes enabler
For the past couple weeks, Twitter has come under attacks that besieged it with more traffic than it could handle. Now comes evidence that the microblogging website is being used to feed the very types of infected machines that took it out of commission.
That's the conclusion of Jose Nazario, the manager of security research at Arbor Networks. On Thursday, he stumbled upon a Twitter account that was being used as part of an improvised update server for computers that are part of a botnet.
The account, which Twitter promptly suspended, issued tweets containing a single line of text that looked indecipherable to the naked eye. Using what's known as a base64 decoder, however, the dispatches pointed to links where infected computers could receive malware updates.
Master command channels used to herd large numbers of infected machines have long been one of the weak links in the botnet trade. Not only do they cost money to maintain, but they can provide tell-tale clues that help law enforcement agents to track down the miscreants running the rogue networks. Bot herders have used ICQ, internet relay chat, and other chat mediums to get around this limitation, but this appears to be the first time Twitter is known to have been employed.
Nazario said he's found at least two other Twitter accounts he suspects were being used in the same fashion, but needs to do additional analysis before he can be sure. The bots using the Twitter account connected using RSS feeds, a technique that allowed them to receive each tweet in real time without the need of an account. It was unclear how many bots connected to the account.
Up to now, the bot designers have done a good job keeping their enterprise under wraps. The original bot software is detected by just 46 percent of the major anti-virus tools, according to this VirusTotal analysis. The updates, which appear to be affiliated with the Buzus trojan, are even stealthier, with only 22 percent of AV engines detecting it.
"This continues the themes that we've been seeing for years now," Nazario told El Reg. "Twitter is just a new mechanism for this, of once you're on the bot pushing out new stuff that's not detected, basically outrunning AV."
Nazario, who offers more details here, says he discovered the Twitter command and control channel by accident, while looking for evidence into denial-of-service attacks that took it out of service for millions of users last week. His finding suggests that in the world of internet crime, large sites can unwittingly be cast as both the victim and the enabler, perhaps simultaneously. ®