Applicability and Validity of AI-Driven Threat Modelling Tools - From Method to Measurement
Author
Bakka, Josephine Marie
Term
4. term
Education
Publication year
2025
Submitted on
2025-05-30
Pages
78
Abstract
Denne afhandling undersøger AI-drevne værktøjer til trusselsmodellering (en struktureret metode til at identificere og prioritere sikkerhedsrisici i software) og vurderer, hvor godt de kan hjælpe og automatisere processen. Med udviklede evalueringsmetrikker og casestudier vurderes en række værktøjers funktioner og hvor pålidelige deres resultater er. Resultaterne viser, at værktøjerne kan øge effektiviteten, men at der stadig er brug for menneskelig vurdering og kontrol. Arbejdet foreslår praktiske evalueringsmetrikker og anbefaler en hybrid human-AI tilgang til sikrere softwareudvikling.
This thesis examines AI-driven threat modeling tools (software that uses artificial intelligence to help identify and prioritize potential security risks in software) and how well they can assist and automate the threat modeling process. Using developed evaluation metrics and case studies, it assesses a set of tools' features and the validity of their outputs. The findings show that these tools improve efficiency, but their results still require human review and oversight. The thesis proposes practical evaluation metrics and supports a hybrid human-AI approach to more secure software development.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Documents