AI-Based Dynamic Risk Scoring Improves Zero Trust Access Control Accuracy and Reduces False Positives
International Journal of Advanced Research in Science Communication and Technology · 19.06.2026 · 76
DOI: 10.48175/ijarsct-37112 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Advanced Research in Science Communication and Technology, online erschienen am 19.06.2026. Verfasst von Abhishek Pal and Dr. Ritu Pal. Die Arbeit nennt 40 Literaturangaben.
Abstract (Originalsprache)
The Zero Trust security model abandons the notion of an implicitly trusted internal network and instead requires every access request to be continuously verified. Conventional Zero Trust deployments, however, rely heavily on static, rule-based policy engines that evaluate requests against fixed conditions. Such rigid rules generalise poorly to evolving threats, produce a large number of false alarms, and impose unnecessary friction on legitimate users. This paper proposes an Artificial Intelligence (AI) based dynamic risk-scoring framework that augments Zero Trust access control with a continuously updated, context-aware risk score. The framework extracts contextual features spanning user identity, device posture, network behaviour, geolocation and temporal patterns, and feeds them to a heterogeneous machine-learning ensemble combining Random Forest, Extreme Gradient Boosting (XGBoost) and a Long Short-Term Memory (LSTM) network. The ensemble outputs a normalised risk score R(t) in the range 0–100, which is compared against adaptive thresholds at the Policy Decision Point to grant, step-up (multi-factor) or deny access. The approach was evaluated on three widely used benchmark datasets—NSL-KDD, CICIDS2017 and UNSW-NB15—and benchmarked against a representative static rule-based baseline. Experimental results show that the proposed model raises detection accuracy to an average of 97.2% and an Area Under the Curve (AUC) of 0.984, while reducing the false positive rate from an average of 17.0% for the static baseline to 3.1%—a relative reduction of more than 80%. These results demonstrate that AI-driven dynamic risk scoring materially improves the accuracy of Zero Trust access decisions while substantially reducing false positives, thereby lowering operational burden without compromising security
Zitieren
APA
Pal, A. P. A. D. R. (2026). AI-Based Dynamic Risk Scoring Improves Zero Trust Access Control Accuracy and Reduces False Positives. International Journal of Advanced Research in Science Communication and Technology, 76. https://doi.org/10.48175/ijarsct-37112
Vancouver
Pal APADR. AI-Based Dynamic Risk Scoring Improves Zero Trust Access Control Accuracy and Reduces False Positives. International Journal of Advanced Research in Science Communication and Technology. 2026:76. doi:10.48175/ijarsct-37112
BibTeX
@article{Pal2026,
author = {Pal, Abhishek Pal and Dr. Ritu},
title = {AI-Based Dynamic Risk Scoring Improves Zero Trust Access Control Accuracy and Reduces False Positives},
journal = {International Journal of Advanced Research in Science Communication and Technology},
year = {2026},
pages = {76},
doi = {10.48175/ijarsct-37112}
}
EUVIMED