Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{210687,
        author = {Sujata Gawade},
        title = {AI-Driven Threat Detection: A Structured Review of Algorithmic, Cloud-Native, and Human-Centered Security Approaches},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {6},
        pages = {1984-1993},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=210687},
        abstract = {The rapid growth of automated, polymorphic, and multi-vector cyber threats has exposed the limitations of traditional signature-based network defense. Artificial intelligence, machine learning, and deep learning have therefore become central to proactive, data-driven threat detection. This paper presents a structured review of AI-driven threat detection across three dimensions: algorithmic adaptability, deployment architecture, and threat surface evolution. 
Three representative studies are examined: an algorithmic taxonomy of support vector machine, random forest, convolutional neural network, and generative adversarial network models; a cloud-native detection architecture evaluated on a simulated enterprise network; and a user-centered framework for AI-generated synthetic threats such as phishing and deepfakes. 
The review compares these studies in terms of architecture, evaluation environment, metrics, operational advantages, and deployment bottlenecks. The analysis identifies four major barriers to production adoption: black-box opacity, latency and resource constraints at the edge, adversarial vulnerabilities including poisoning and evasion, and false-positive-driven alert fatigue. 
Finally, the paper outlines future research directions, including neuro-symbolic AI, adversarial-aware model compression, automated adversarial retraining loops, and human-in-the-loop explainable interfaces. The central finding is that production-ready AI-driven threat detection requires not only accurate models, but also transparent, scalable, adversarial robust, and human-aware defensive architectures.},
        keywords = {AI-driven threat detection, machine learning, deep learning, explainable AI, adversarial machine learning, cloud-native security, cybersecurity, human-in-the-loop security.},
        month = {November},
        }