University of Twente Student Theses

Login
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.

Detecting BOLA Vulnerabilities with Large Language Models

Johansens, E. (2025) Detecting BOLA Vulnerabilities with Large Language Models.

[img] PDF
431kB
Abstract:Broken Object Level Authorization (BOLA) is widely recognized as one of the most critical API security risks. Since its initial inclusion in the 2019 publication, the vulnerability has retained the number-one ranking in the OWASP 2023 API Security Top Ten. Detecting BOLA attacks manually is labour-intensive and error-prone, and the existing automated tools do not provide full coverage over every API. The paper investigates whether large language models (LLMs) can effectively identify BOLA vulnerabilities in REST APIs. The research first presents a dataset of 12 REST APIs, described in the OpenAPI 3.0 format. A prompt engineering approach is then employed by giving the LLM a context-rich and role-specific prompt and asking it to identify BOLA vulnerabilities. Four state-of-the-art LLMs are evaluated using the dataset, and their outputs are compared against the ground truth. The results show that LLMs achieve high accuracy and recall but suffer from low precision, producing many false positives. Model performance is compared against each other, and the Deepseek-R1 model achieves the best overall performance. Lastly, small-parameter LLMs are explored; however, the output shows a fundamental lack of knowledge in cybersecurity.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:50 technical science in general, 54 computer science
Programme:Business & IT BSc (56066)
Link to this item:https://purl.utwente.nl/essays/107423
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page