Attackers have set up around 10,000 malicious subdomains on accounts belonging to GoDaddy customers, according to a new report from Cisco's Talos Security Intelligence and Research Group.
The technique, called domain shadowing, first appeared in 2011 but has dramatically grown in popularity in the past three months in connection with the Angler Exploit Kit.
"Basically, it looks like they were prototyping it in 2011, then decided to go live with it," said Craig Williams, security outreach manager at Talos.
Williams added that he expects other exploit merchants to soon offer similar functionality because of the competitive nature of the malware marketplace ever since the previous best-selling exploit kit has been taken down by authorities.
"Paunch was put in jail last year and the Blackhole exploit kit basically disappeared from the market," he said. "Now everyone is trying to capture that market share. Overall, it seems that the exploit kit authors have really been racing to refine their exploit kits."
The way domain shadowing works is that the attackers break into GoDaddy accounts of legitimate customers, get access to the domain name registrations for their domains, then set up as many subdomains as they want.
Since the attackers do not touch the main domain name, the victims may not realize that they've been hijacked.
"Unless they have a particular reason to do so, people rarely log into their domain registrar," Williams said. "They may never even know their account has been compromised."
GoDaddy was involved in all of the account hijackings.
"There was only one outlier," said Talos security researcher Nick Biasini. "There was someone who had domains under multiple accounts, with GoDaddy and also with another provider."
GoDaddy is an attractive target because it is the single biggest registrar on the Internet, accounting for almost a third of all domains.
Cisco has identified around 10,000 malicious subdomains registered from several hundred GoDaddy accounts, according to Biasini.
Around 1,500 of those domains were registered back in 2011. But the bulk -- more than 7,500 -- have appeared in the past three months, with February seeing more activity than January, which itself was an increase over December.
The number of compromised user accounts has been growing, as well.
Attackers use common dictionary words to create some subdomains, then random strings of characters to create others. The former are used for redirection pages, and the latter, of which there are significantly greater numbers, for malicious landing pages. Attackers cycle through different landing pages very quickly and, moreover, also cycle through different IP addresses for those pages.
"The end goal for the bad guys is to have so many domains and use each for such a short period of time that standard blacklisting techniques are not effective," said Williams.
Security systems need to use other techniques as well, to detect these kinds of attacks, including behavior-based approaches, sandboxing, and analyzing the actual files that the sites attempt to install.
Domain shadowing also hinders researchers, Williams said.
"It's easy for us, because we're Cisco, and we have this gigantic cloud infrastructure," he said. "But for anybody who doesn't have this ability, by the time they detect that a piece of malware has been downloaded, they have no way to figure out what happened because it's already moved on to a different website."
According to the report, the attackers have only made use of about a third of the domains so far, indicating that they have plenty of domains in reserve.