Zeus Attackers Deploy Honeypot Against Researchers, Competitors

Attackers turned the tables on both their competitors and researchers investigating a recent Zeus attack, which targeted quarterly federal tax payers who file electronically, by feeding them a phony administrative panel with fake statistics.

The massive and relatively sophisticated spam campaign last month posed as email alerts to victims, notifying them that their electronic federal tax payments had failed and sending them to a link that both infects the victim with the Zeus Trojan and sends victims to the legitimate Treasury Department website, eftps.gov, for filing quarterly taxes.

Brett Stone-Gross, a researcher with The Last Line of Defense, discovered that attackers had set up a ruse for those trying to hack or access its administrative interface for the malware after studying the back-end malware server used in the EFTPS attack. The purpose appeared to be all about providing false information. Stone-Gross says the toolkit used in the attack came with an administrative interface that acts as a hacker's honeypot of sorts, gathering intelligence about the researchers or other users who try to access the console login or hack into it.

The login system to the "admin panel" practically begs to be hacked: It accepts default and easily guessed passwords as well as common SQL injection strings, according to Stone-Gross.

Most exploit toolkits come with an admin interface that manages exploits and payloads, and tracks exploit success rates, but this fake one was a new twist, Stone-Gross says. He found the fake panel while browsing the gang's source code. "It had a directory called 'fake admin' where they stored the logs of all of the IP addresses of people who tried the console and tried to access it," Stone-Gross says. There were also comments in Russian, he says.

"The faked admin panel serves two purposes: leading the researchers looking at their infrastructure, and they want to see who their competitors are," he says. They can then blacklist the researchers or use the information to DDoS or attack security vendors trying to investigate their malware campaigns, he says.

Joe Levy, CTO at Solera Networks, one of the first researchers to spot the EFTPS attack last month, says his team didn't see any honeypots during their investigation, but the appearance of such traps aren't surprising. "It is well-known that we are in an evolutionary arms race with cybercriminals. We've seen such signs of maturation as copycat malware, cybercrime ring wars, and even collusion and consolidation," Levy says. "Naturally, they have used and will continue to use honeypots for all the same reasons that the research community uses them: to better learn about their adversaries, as a tactic of deception, and to spread misinformation and uncertainty ... We need to keep mindful of this, but we can't permit such deceit to stifle or thwart our progress."

The attackers also offered what they advertised as "new botnet" malware, which rather than providing a peek at their next-generation bot instead gave the attackers a way to gauge what their competitors were up to, such as launching a fake AV campaign, Stone-Gross says.

Thorsten Holz, senior threat analyst at LastLine and assistant professor of computer science at Germany's Ruhr-University Bochum, says he thinks the "new botnet" button could have been for collecting new samples. "As an attacker, I would also love to learn what competitors would like to install on my infected machines. Brett found that the attackers logged a lot of information related to the login attempts, maybe to track more closely who wants to hack their back end," he says.

This helps them glean other details about their visitors, such as the browser version being used by the researcher and other "fingerprints" of their software, Stone-Gross says.

And the statistics on infected machines the attackers provided on their fake admin console were inflated. Stone-Gross was able to get a peek at their real database for brief intervals to see the real numbers, and they were far lower, he says.

Such anti-forensics activity by the bad guys is likely to become more common in the future, the researchers say. This, of course, poses problems for researchers and investigators trying to get a handle on the number of infections or the sizes of botnets. "Measuring the actual size of a botnet is already hard; it now gets even harder since we cannot trust the logging data without analyzing it in detail," Holz says.

Any data accessible via a Web interface should be considered suspect, says Stone-Gross, who blogged about his findings today. "Unless you have the back-end source code, it's hard to say what's really going on," he says.

Paul Henry, security and forensic analyst at Lumension, says disinformation tactics will continue. "In some respects, this smells like a marketing tactic by the malware author. It is important to remember that malware is big business today," Henry says. "We can expect that they will adopt marketing tactics that legitimate software providers have used historically to sell their wares."

Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.