In Episode S7E2, host Steven Bowcut and Edward Wu, founder and CEO of Dropzone AI, discuss the use of AI in cybersecurity, specifically focusing on the concept of agentic AI. Edward explains that agentic AI refers to AI systems that can autonomously perform complex tasks without incremental instruction from users. He emphasizes the importance of coachability in AI systems, likening them to digital workers who need to adapt to an organization’s specific needs.
Summary
In this podcast conversation, Steven and Edward explore the potential of AI in cyber defense, emphasizing its role in augmenting human security teams. Edward highlights a case study where AI enables a tech startup to function as if it has more engineers on staff. They delve into the concept of ‘agentic AI’ and its application in coaching AI systems, with Edward noting Dropzone AI’s recognition as a ‘cool vendor’ by Gartner in October of the previous year—an important milestone for the company.
Edward explains that agentic AI refers to systems capable of autonomously performing complex tasks without incremental instructions from users. He underscores the importance of coachability in AI, comparing these systems to digital workers who must adapt to an organization’s specific needs. Steven adds that the value of a team member grows exponentially as they learn the operational nuances of the organization. Edward points out the trend of utilizing agentic AI to enhance productivity by offloading tedious tasks, such as tier-one analytical work.
The duo also discusses the ethical training of large language models, addressing the challenges related to using unlicensed and private customer data. Edward raises concerns about the risks involved and advocates for responsible data handling, urging vendors to keep a clear distinction between private information and data used for system improvement. Steven expresses worry about the potential for agentic AI systems to learn from humans who may not fully grasp ethical standards. In response, Edward emphasizes the significance of case studies from fields like medicine.
Finally, they discuss the growing adoption of AI in cybersecurity, with Edward noting that the technology has matured significantly in the past year. He highlights the potential for attackers to exploit large language models as well. Sharing his vision for the future, Edward aspires to create the most capable and trustworthy AI security analyst, which would empower organizations of all sizes to investigate security alerts promptly, making it harder for attackers to succeed. Steven conveys enthusiasm about the prospect of using digital workers as a force multiplier for startups and small businesses.
About our Guest
Edward is the founder and CEO of Dropzone AI. Dropzone AI is the first AI SOC analyst that autonomously investigates alerts 24/7. Edward is an expert in applied AI/ML for cybersecurity and next-gen cyber defense. He previously built the network attack detection product at ExtraHop Networks and researched automated binary analysis and software defenses at UW Seattle and UC Berkeley. Edward holds 30+ patents in ML and cybersecurity and is a contributor to the MITRE ATT&CK framework.
This episode is a must-listen for anyone interested in agentic AI and digital workers.
Click the image below to listen to this episode of Brilliance Security Magazine Podcast.
Steven Bowcut is an award-winning journalist covering cyber and physical security. He is an editor and writer for Brilliance Security Magazine as well as other security and non-security online publications. Follow and connect with Steve on Twitter, Instagram, and LinkedIn.