A Taxonomic Classification of Insider Threats: Existing Techniques, Future Directions & Recommendations

In the last two decades, the number of rapidly increasing cyber incidents (i.e., data theft and privacy breaches) shows that it is becoming enormously difficult for conventional defense mechanisms and architectures to neutralize modern cyber threats in a real-time situation. Disgruntled and rouge employees/agents and intrusive applications are two notorious classes of such modern threats, referred to as Insider Threats, which lead to data theft and privacy breaches. To counter such state-of-the-art threats, modern defense mechanisms require the incorporation of active threat analytics to proactively detect and mitigate any malicious intent at the employee or application level. Existing solutions to these problems intensively rely on co-relation, distance-based risk metrics, and human judgment. Especially when humans are kept in the loop for access-control policy-related decision-making against advanced persistent threats. As a consequence, the situation can escalate and lead to privacy/data breaches in case of insider threats. To confront such challenges, the security community has been striving to identify anomalous intent for advanced behavioral anomaly detection and auto-resiliency (the ability to deter an ongoing threat by policy tuning). Towards this dimension, we aim to review the literature in this domain and evaluate the effectiveness of existing approaches per our proposed criteria. According to our knowledge, this is one of the first endeavors toward developing evaluation-based standards to assess the effectiveness of relevant approaches in this domain while considering insider employees and intrusive applications simultaneously. There have been efforts in literature towards describing and understanding insider threats in general. However, none have addressed the detection and deterrence element in its entirety, hence making our contribution one of a kind. Towards the end of this article, we enlist and discuss the existing data sets. The data sets can help understand the attributes that play crucial roles in insider threat detection. In addition, they can be beneficial for testing the newly designed security solutions in this domain. We also present recommendations for establishing a baseline standard for analyzing insider-threat data sets. This baseline standard could be used in the future to design resilient architectures and provide a road map for organizations to enhance their defense capabilities against insider threats.