Open Access

REVIEW

Phishing, Vulnerabilities, and AI Defense: A Systematic Review of Cybersecurity Challenges and GRU-Based Mitigation Strategies in Digital Microfinance Institutions

Richard Mathenge^*, Catherine Mukunga, Ephantus Mwangi

School of Pure and Applied Sciences, Kirinyaga University, Kerugoya, Kenya

* Corresponding Author: Richard Mathenge. Email:

Journal of Cyber Security 2026, 8, 129-151. https://doi.org/10.32604/jcs.2026.077183

Received 04 December 2025; Accepted 27 January 2026; Issue published 11 March 2026

Abstract

The rapid digitization of microfinance institutions (MFIs) has strengthened financial inclusion but has simultaneously increased exposure to phishing attacks and other cybersecurity threats driven by organizational, technical, and human vulnerabilities. Grounded in socio-technical systems theory, this systematic analysis evaluates AI-based mitigation strategies, with particular emphasis on gated recurrent unit (GRU) architectures. It compares them with Transformer and LSTM models. GRUs are prioritized due to their computational efficiency and suitability for low-resource environments typical of digital MFIs. Following PRISMA 2020 guidelines, 32 empirical studies published between January 2012 and April 2025 were analyzed from the Web of Science, ScienceDirect, Google Scholar, IEEE Xplore, and Scopus. Thematic synthesis was conducted using the Braun and Clarke framework, and methodological quality was assessed using the Mixed Methods Appraisal Tool (MMAT). The findings indicate that obsolete infrastructure, limited employee awareness, and weak governance structures account for approximately 67% of cybersecurity incidents in MFIs. Under experimental conditions, GRU-based models achieved phishing-detection accuracies of 92% to 96%, demonstrating strong performance in sequential behavior analysis. Despite these advantages, deployment remains constrained by infrastructural limitations, limited model explainability, and scarcity of domain-specific datasets. This study proposes an implementation roadmap integrating explainable AI, ethical governance, and region-specific capacity building, alongside a vulnerability-solution matrix linking threat vectors to appropriate AI-based countermeasures. The findings provide a structured foundation for developing secure, scalable, and AI-enabled digital financial ecosystems for legislators, cybersecurity practitioners, and MFI stakeholders.

---

Keywords

---