Crimeware Developers Shift To More Obfuscation, Java Exploits

When Microsoft patched a vulnerability in the Windows XML Core Services this week, malware authors had already been exploiting the flaw in successful attacks through Internet Explorer, most notably as part of the Blackhole exploit kit.

While the authors of that exploit kit had focused last year on making the code harder to reverse-engineer, more recently they have bulked up the software's store of useful exploits, says Jason Jones, lead for the advanced security intelligence team at Hewlett-Packard's DVLabs, who will present his research on exploit kits at the Black Hat USA security conference later this month. In the past year, the authors of the Blackhole kit have augmented the program with a handful of successful attacks against Java vulnerabilities, as well as those in Internet Explorer.

"The authors of these kits are seeing what works well, and then they work to improve it, and they also work really hard at trying to keep their code out of security researchers' hands at the same time," Jones says. "These people take their jobs fairly seriously, and they also want to keep making money."

The focus on adding more exploits, especially Java exploits, has been a common trend among other such kits, he says. While the focus on compromising systems through the XML Core Services vulnerability is a departure from that, other toolkits -- such as the Phoenix toolkit and several programs being produced by Chinese groups -- have also focused on Java, he says.

[ Recent widespread spam runs posing as convincing-looking email messages from high-profile organizations are all part of a single, orchestrated attack campaign using the Blackhole exploit kit. See Series Of Convincing Spam Runs Part Of One Massive Advanced Attack Campaign. ]

The group behind Blackhole started ramping up its focus on Java following the success of an earlier exploit included as part of the toolkit last year.

"The Java exploit that they included in the kit last year had such a high success rate, in the 70- to 80-percent range, they saw that as, 'Hey, this is great. People will keep buying this if we can add more like this,'" he says.

The focus on incorporating Java into exploit kits started in the third quarter of 2010, according to research (PDF) presented by consultant Dan Guido, now CEO at security start-up Trail of Bits. In a study of the 15 most popular toolkits, Guido found that 11 out of 15 had at least one Java exploit and two-thirds of those had at least two Java exploits.

Because legacy versions of the Java runtime usually remain on a computer even after an update, attackers have had good success against the ubiquitous software. Yet the latest XML exploit shows that the kit creators are willing to do a bit more work to get a jump on the defenders, as well. It would not have taken malware developers long to create the attack, especially because a security researcher had produced a recipe for the exploit in the form of pseudo-code that had been posted online.

"It's the advances in technical skill in exploit development skills that I really care about," Guido says. "The Java thing is a tiny, tiny example that these guys are willing to do more than copy and paste. These guys are willing to sit down and type out 50 lines of Java, which is not a lot, but it's more than they were willing to do last year."

Because of the focus on exploits for Java vulnerabilities, HP's Jones recommends that companies turn off Java on computers that do not need to run the software. In his previous research, Guido found that disallowing Java in the Internet zone would protect against the exploits in the existing toolkits.

While advances in obfuscation are interesting, the attacks have no chance to escape eventual notice, Guido says. Of course, they don't have to, he says.

"They don't need to stay stealthy," he says. "What they need to do is delay the amount of time between when it hits an AV analyst's desk to when the company pushed out a detection for the attack, so that those cycles that they go through -- new attack, new defense, new attack, new defense -- is as long as possible."

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