8 Feb
2024
8
min read

How can we leverage LLM's to aid vulnerability management?

There is a wealth of opportunity when using LLM's (Large Language Models) to aid vulnerability management and even vulnerability detection. In this blog, we'll discuss how LLM's can support your vulnerability process and which key areas you should focus on when applying LLM's to your technology stack.

Sian-Louise Montgomery
Marketing Consultant

Using LLM's (Large Language Models) can support vulnerability management and detection through multiple stages of the technology environment.

Firstly, LLM's can be applied to identifying where vulnerabilities are being introduced in systems to begin with. Here, we can look at things like tickets and pull requests and being able to process that information in terms of what kinds of vulnerabilities are likely to have come from. The changes that have been detected are an incredibly valuable resource to any security testing programme.

Secondly, LLM's can take feeds of information from multiple different vulnerability scanners. This means that LLM's are able to say "Okay, these two pieces of information from SAST and DAST..." or from a cloud configuration reviewer and from a network scanner, and see that these two things relate to the same vulnerability. Therefore, they should be combined together.

Thirdly, LLM's can support prioritisation in a big way. Being able to look at vulnerabilities across the estate, but in the context of the estate being able to interpret things like the technology stack and the vulnerabilities holistically. Then, we can say "Okay, businesses that have these problems are likely to have these particular vulnerabilities pop up". Or even, "We know that businesses have been able to approach things in this way to make material changes to themselves" and then looking at that data in a more strategic way is a massive opportunity.

All of this can be done without the use of LLM's, you can do all of that in the traditional, manual way when you're talking about just a few vulnerabilities or even a few hundred. But, for most organisations, we're talking 10 to 200,000 vulnerabilities in a backlog that they're trying to deal with. Every new scanner and every new tool that they think is incredibly valuable to them is likely just adding to that backlog.

So, unless you've got something like an LLM which is really able to handle data at that kind of scale, you're not going to be able to address the vulnerability management problem as effectively.

bug report

Get a Free Trial  From Cytix

Haven’t tried Cytix yet? Try our free trial to see how it works.

Get a Free Trial

Start Detecting Vulnerabilities Others Miss Today

  • Detect Vulnerabilities Faster
  • Patch Vulnerabilities Faster
  • Be more compliant
Learn More
business

Detect, Resolve &
Patch Faster With Cytix

Get a free test today and see how it works.
CTA Image
cta rectangle image
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.