Black Hat USA 2014: Third-Party Vulns Spread Like Diseases

As security professionals grapple with how to classify and triage vulnerabilities in third-party libraries and components their software depends upon, it may help to think about the spread of these vulnerabilities the way a public health professional views the spread of infectious disease. Two researchers at Black Hat USA next month plan to present data that shows how the spread of attack surface from something like a Heartbleed vulnerability instance looks very similar to an epidemiological event, like a widespread flu outbreak.

The idea came from Kymberlee Price, director of ecosystem strategy at Synack, who had a background in public health before her time as a security pro focusing on red team management and vulnerability metrics analysis. She teamed up with Jake Kouns, CEO of the Open Security Foundation and one of the executives who help to oversee OSF's Open Sourced Vulnerability Database (OSVDB.org) to examine OSVDB data and see how well her hunch held up. The numbers showed that, while libraries and other third-party components offer a huge leg up in speed and efficiency of development, the practice greatly amplifies the impact of a single vulnerability in any one of these components.

"We're not suggesting that people shouldn't use third-party libraries, because they have a valid purpose in the development," Price says. "We're going to talk about what you as a developer or an IT manager need to be doing or thinking to protect your customers or your profits because you're ultimately accountable for all of the code you ship -- whether you wrote it or not."

The greater the success of the library, the greater the threat posed by a vulnerability in it. That's because the more software third-party components touch, the more potential they have to spread the pestilence when vulnerabilities are present.

"We're going to look at a number of widely used third-party libraries in terms of how many vulns they have, how many products they're used in, and what the relative severity is of those," Price says. "So it's not just a lot of low-level hygiene fixes that you'll have to worry about if you use this library, but these are the big advisory-class problems and how much you should be worried about them."

Price and Kouns will unveil some of the hard numbers in their talk to unveil which libraries are some of the biggest offenders in expanding attack surfaces. But as they preview their talk, they're not afraid to give an early peek at their biggest message. That is the persistent truth that the kind of epidemiological spread presented by third-party vulnerabilities means that organizations can't simply depend on CVSS to guide their prioritization of risk when it comes to these bugs. Even when a bug in a third-party component is not necessarily a high CVSS vulnerability, it could present quite a bit of threat, due to pervasiveness and how code is deployed.

"I think it's fair to talk about Heartbleed as a case study because everyone's pretty familiar with that at this point," Price says. "If you look at the CVE associated with the Heartbleed vulnerability, you'll find over 200 advisories tied just to that one CVE, different products by different vendors. While that CVSS score on that vuln is only a 5.0, once you start looking at what's being accessed in some of these products it starts looking significantly more impactful. It may not be a 10.0, but it's incredibly damaging."

This lesson is especially important for software vendors to digest as they reexamine their incident response and maintenance practices, a topic which Price hopes to expand on during the talk. "So one of the highlights of the talk is, you need a robust incident response plan, but you also need a proactive maintenance plan, so you're not only updating on the emergency events," she says, explaining that if a vendor has no maintenance plan in place and a library has an impactful vulnerability, it will be significantly more difficult to make a fix if that vendor has missed four or five updates prior to the patch in question. "The problem when a catastrophic vulnerability occurs and you're four versions old is that there have been so many incremental changes to the library that it'll take major architectural changes in the product to jump to the right version."

Meanwhile, enterprises should also be paying close attention to these vulnerabilities -- even the low-CVSS-scored bugs -- so they can turn up the heat on vendors to make fixes quickly and ask for workarounds until patches are available.

"One of the things I think is important for IT admins to do is, first off, know what's in their environment. Not just what products are in there, but what third-party code is in those products," she says, explaining that this will allow customers to put the screws to vendors when news of big third-party vulnerabilities surface. "Believe it or not, vendors care what customers have to say -- and if customers are saying 'Security is a priority to me,' then security becomes a priority for the vendor."