Artificial Intelligence in Cybersecurity: Detecting and Preventing AI-Powered Cyberattacks
Algorithm Scientific Research Journal · 20.03.2026 · 107-124
DOI: 10.61438/asrj.v1i1.331 ↗Kurz gefasst
Fachartikel in der Zeitschrift Algorithm Scientific Research Journal, online erschienen am 20.03.2026. Verfasst von Mohammad Dawood Raqib. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
Artificial Intelligence (AI) serves as a foundational driver in reshaping modern digital industries, yet its widespread deployment introduces structural vulnerabilities that cybercriminals increasingly exploit. This study provides a comprehensive exploration of contemporary AI-powered cyberattack methodologies and establishes a unified defensive metadata framework to detect and counteract these automated vectors. We analyze the execution constraints of Adversarial Machine Learning (AML) through white-box perturbations and black-box surrogate classifier transferability, alongside training-phase risks including dataset poisoning and clean-label backdoor sample injections. To mitigate these threats, this paper evaluates an empirical pipeline combining multi-classifier stacked ensembles (Super Learners) with unsupervised generative adversarial architectures (GANomaly) for latent-space network telemetry compression. Experimental validation conducted across benchmark distributions—including NSL-KDD, Drebin-215, and Edge-IIoTset—demonstrates that gradient-boosted stacking frameworks significantly outperform standalone classifiers by minimizing prediction variance and eliminating local coordinate blind spots. Finally, this paper addresses the core architectural friction domains encountered during enterprise integration, highlighting how structured human-AI teaming cultivates long-term, sustainable resilience within automated industrial control systems.
Zitieren
APA
Raqib, M. D. (2026). Artificial Intelligence in Cybersecurity: Detecting and Preventing AI-Powered Cyberattacks. Algorithm Scientific Research Journal, 107-124. https://doi.org/10.61438/asrj.v1i1.331
Vancouver
Raqib MD. Artificial Intelligence in Cybersecurity: Detecting and Preventing AI-Powered Cyberattacks. Algorithm Scientific Research Journal. 2026:107-124. doi:10.61438/asrj.v1i1.331
BibTeX
@article{Raqib2026,
author = {Raqib, Mohammad Dawood},
title = {Artificial Intelligence in Cybersecurity: Detecting and Preventing AI-Powered Cyberattacks},
journal = {Algorithm Scientific Research Journal},
year = {2026},
pages = {107--124},
doi = {10.61438/asrj.v1i1.331}
}
EUVIMED