Analyzing and Mitigating Prompt Injection Attacks in AI-driven Cloud Applications

Badgujar, Shreyas Kiran (2025) Analyzing and Mitigating Prompt Injection Attacks in AI-driven Cloud Applications. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (870kB) | Preview
Preview	PDF (Configuration Manual) Download (719kB) | Preview

Abstract

Large Language Models (LLMs) are becoming more directly involved with interactive applications, but they are susceptible to prompt injection attacks that influence system response. The current literature primarily categorizes attacks or introduces single-layer defences, which provide less protection in real-time. In this study, a multi-stage defence pipeline which mix the TF-IDF and Logistic Regression–based classifier, Structured Query Filtering, prompt sanitisation, and the controlled GPT-4 processing is built to confirm safe and consistent handling of the user inputs. The Experimental results show 94.6% accuracy, 92.3% precision, and 93.1% recall, determining the strong recognition ability. The model improves the secure implementation of LLM-powered applications by ensuring secure, flexible and efficient protection against changing threats of prompt injection.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Prior, Michael UNSPECIFIED
Uncontrolled Keywords:	Prompt Injection; LLM Security; Zero-Shot Detection; Structured Query Filtering; GPT-4
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security B Philosophy. Psychology. Religion > BJ Ethics > Conduct of life > Reliability > Information integrity > Data integrity
Divisions:	School of Computing > Master of Science in Cyber Security
Depositing User:	Ciara O'Brien
Date Deposited:	03 Sep 2026 10:17
Last Modified:	03 Sep 2026 10:17
URI:	https://norma.ncirl.ie/id/eprint/9789

Actions (login required)

View Item