Towards leveraging agentic AI for managing security and usability trade-offs
Rathee, Anuj (2025)
Kandidaatintyö
2025
School of Engineering Science, Tietotekniikka
Kaikki oikeudet pidätetään.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20251209116409
https://urn.fi/URN:NBN:fi-fe20251209116409
Tiivistelmä
This bachelor's thesis explores how an AI agent can help manage security usability trade-offs between security and usability in different scenarios. The research addresses the limitations of traditional, static security mechanisms that often cause friction and are inefficient in dynamic, modern contexts.
The primary objective of the thesis was to develop a theoretical framework for an AI agent capable of making balanced, intelligent, explainable decisions. This was achieved by introducing the Agentic Trade-off Decision Model (ATDM), which utilizes utility-based reasoning combined with contextual risk scoring and dynamic weighting to formally balance security and usability based on the situation.
The thesis also had an objective to evaluate user perception of this adaptive security behavior. This was investigated through an empirical user study using scenario-based evaluations. The findings show that the ATDM approach, supported by a clear Explainability Mechanism, significantly improved user perception, resulting in lower cognitive load, higher trust, and better perceived usability compared to static systems.
The study concludes that adaptive, context-aware mechanisms offer a more effective and user-centric approach to managing security trade-offs. The findings strongly support the use of utility-based agentic models as a systematic way to deliver necessary security while minimizing unnecessary interruptions, thereby increasing user compliance and overall trust.
The primary objective of the thesis was to develop a theoretical framework for an AI agent capable of making balanced, intelligent, explainable decisions. This was achieved by introducing the Agentic Trade-off Decision Model (ATDM), which utilizes utility-based reasoning combined with contextual risk scoring and dynamic weighting to formally balance security and usability based on the situation.
The thesis also had an objective to evaluate user perception of this adaptive security behavior. This was investigated through an empirical user study using scenario-based evaluations. The findings show that the ATDM approach, supported by a clear Explainability Mechanism, significantly improved user perception, resulting in lower cognitive load, higher trust, and better perceived usability compared to static systems.
The study concludes that adaptive, context-aware mechanisms offer a more effective and user-centric approach to managing security trade-offs. The findings strongly support the use of utility-based agentic models as a systematic way to deliver necessary security while minimizing unnecessary interruptions, thereby increasing user compliance and overall trust.