Discovering Software Vulnerabilities with LLMs
Dnr:
NAISS 2026/4-49
Type:
NAISS Small
Principal Investigator:
Emre Süren
Affiliation:
Kungliga Tekniska högskolan
Start Date:
2026-01-09
End Date:
2027-01-01
Primary Classification:
10201: Computer Sciences
Allocation
Abstract
This research seeks to investigate three distinct applications: the utilization of open-source LLMs for the automated detection and exploitation of conventional software vulnerabilities, the autonomous discovery of web vulnerabilities by AI agents, and the identification of flaws in novel-generation LLM-powered applications.