Development of an AI-Driven Behaviour Analytics Model for Predicting Insider Threat in Organizational Networks 04 November 2026
Year: 2026
Type: article
Abstract: Insider threats represent one of the most complex challenges in contemporary enterprise cybersecurity due to the legitimate credentials and elevated access privileges held by internal actors. Traditional perimeter-focused security mechanisms, including firewall policies and signature-based Intrusion Detection Systems (IDS), fail to detect malicious or inadvertent activities initiated from authorized endpoints. This paper presents a hybrid AI-driven User and Entity Behavior Analytics (UEBA) framework designed to predict and isolate insider threats across complex organizational networks. The proposed architecture integrates continuous multi-source data ingestion—comprising NetFlow records, system authentication logs, file interaction frequencies, and temporal activity matrices—into a dual-stage modeling pipeline. A recurrent sequence autoencoder captures baseline behavioral drift across temporal windows, while an Isolation Forest classifier identifies fine-grained anomalous deviations. Evaluated against enterprise benchmark telemetry, the model achieves a detection accuracy of 94.7% with a false positive rate of 2.1%, significantly outperforming static threshold models while mitigating Security Operations Center (SOC) alert fatigue.
Authors: SWAMADU, SURAIT, niyonkuru, Steven
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225896768
