Zsolt-Levente Kucsván

I am a security researcher and PhD student in Computer Science at Babeș-Bolyai University, in Romania. My research is about the safety of agentic AI for automation.

Previously, I was a PhD student at the University of Twente, in the Netherlands, where I worked on automating incident response in Security Operation Centers (SOCs).

Security operators rely on playbooks, step-by-step procedures that tell them how to contain and mitigate a threat. Writing and maintaining these by hand does not scale. My current work looks at generating playbooks automatically, for known threats and for new ones, from sources such as Cyber Threat Intelligence reports.

To do this I study how Large Language Models (LLMs) can help build defensive tooling. Because LLMs make mistakes and hallucinate, a large part of my work is about measuring how reliable they are and when it is safe to trust their output. My current research builds on this question, extending it to AI agents that act autonomously.

Before my PhD studies, I earned an MSc in Cyber Security through the EIT Digital double degree programme at the University of Trento and the University of Rennes 1, and a BSc in Computer Science at Sapientia Hungarian University of Transylvania. Outside research I play CTFs with the Twente Hacking Squad.

Research interests: safety of agentic AI · security operations & incident response · LLMs for security · cyber threat intelligence · ML for network security.