International Journal of Computer Techniques Volume 12 Issue 4 | The Insider Risk of Artificial Intelligence in Financial System through the Lens of Large Language Models
The Insider Risk of Artificial Intelligence in Financial Systems through the Lens of Large Language Models
Authors:
Ishrak Alim, M.S. in Accounting Analytics, University of New Haven, Connecticut, USA (alimishrak@gmail.com)
Tasnia Farzana Matin, M.S. in Digital Marketing Analytics, Montclair State University, New Jersey, USA (matinfarzana77@gmail.com)
Takib Md Masudul Hasan Prodhan, Account Executive, T&S Buttons Lanka Ltd, Dhaka, Bangladesh (bd.takib1@gmail.com)
Md Lahaduzzaman Lahad, Digital Marketing Executive, Cloud Bridge Consultancy, Dhaka, Bangladesh (lahaduzzaman.lahad@yahoo.com)
Journal: International Journal of Computer Techniques (IJCT)
Volume: 12 | Issue: 4 | Publication Date: July – August 2025
ISSN: 2394-2231 | Journal URL: https://ijctjournal.org/
Abstract
This paper examines insider risks introduced by Large Language Models (LLMs) in financial systems. It presents a taxonomy of vulnerabilities including prompt injection, data leakage, and role-boundary breaches. A structured risk matrix and mitigation strategies are proposed, covering technical controls and governance frameworks. Real-world use cases illustrate the need for explainable AI, proactive monitoring, and ethical deployment to safeguard financial workflows.
Keywords
Large Language Models (LLMs), Insider Threats, Financial Systems Security, AI Governance, Prompt Injection, Data Leakage, Role-Based Access Control, Model Auditing, Explainable AI (XAI), Risk Assessment Matrix
Conclusion
LLMs offer transformative capabilities in finance but introduce nuanced insider risks. Role-based prompt isolation, session logging, and governance dashboards are essential for secure deployment. The paper emphasizes the need for continuous auditing, ethical AI adoption, and domain-specific model calibration to ensure trust and resilience in financial ecosystems.
References
Includes 30+ references from IEEE, arXiv, Springer, SSRN, and academic journals covering LLM vulnerabilities, financial AI applications, insider threat detection, and governance frameworks.
