Using LLMs to Automate and Streamline Cyber Threat Analysis

TechSpective Podcast Episode 122

There are a number of steps involved in cyber threat analysis to review event information and determine which events are benign or innocuous and which are malicious–or at least deserve greater scrutiny. For the most part, it’s not that it is exceptionally complex or difficult work. The issue is mostly a matter of scale. The sheer volume of security events generated makes the initial triage to flag the ones that need more attention and collect the peripheral information to provide the context necessary for effective investigation is a Sisyphean task.

It is an ideal scenario for employing AI. Machine Learning and Artificial Intelligence have been used for years to automate and streamline these sorts of tasks. Now, generative AI built on LLMs (Large Language Models) is transforming cybersecurity analysis. Anything that is routine and repeatable and can be clearly-defined should be automated with AI.

Adobe Security has developed a set of tools dubbed “Project Sherlock” to do just that. Tiberiu Boros and Radu Chivereanu, machine language engineers with Adobe Security, join the podcast to talk about ‘Project Sherlock’–an internal Adobe project utilizing generative AI to streamline cyber threat analysis and incident response.

Check out the full episode for more on Adobe’s “Project Sherlock.” We also discuss the challenges facing organizations and developers around the volume and relevance of the data sets used to train large language models, and some of the issues with AI hallucination and the need to validate the responses provided from generative AI models.

The podcast itself is audio-only, but the video of our conversation is also available on YouTube if you prefer:

Please ask questions and share your thoughts on the topic in the comments below. Also, please subscribe to the TechSpective Podcast through your favorite podcast platform and share the podcast with your peers and friends.

If you enjoy the podcast, I would also be grateful if you could take 2 minutes to rate and review the podcast on iTunes or wherever you listen.

About
Latest Posts

Tony Bradley

Editor-in-Chief at TechSpective

I have a passion for technology and gadgets--with a focus on Microsoft and security--and a desire to help others understand how technology can affect or improve their lives. I also love spending time with my wife, 7 kids, 4 dogs, 7 cats, a pot-bellied pig, and sulcata tortoise, and I like to think I enjoy reading and golf even though I never find time for either. You can contact me directly at tony@xpective.net. For more from me, you can follow me on Threads, Facebook, Instagram and LinkedIn.

TechSpective Podcast Episode 122

Related Posts

4 thoughts on “Using LLMs to Automate and Streamline Cyber Threat Analysis”