What does this research mean for the field?

While hybrid AI models outperform standalone methods in cybersecurity applications, their real-world deployment is severely hindered by high computational costs, lack of interpretability, and an over-reliance on academic benchmark datasets. Novelty: ClaimNovelty.SYNTHESIS. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To evaluate artificial intelligence applications in cybersecurity, focusing on intrusion detection, malware analysis, and the divergence between benchmark laboratory results and operational deployments.

September 14, 2026Iraqi Journal for Computers and InformaticsOpen Access

AI-Powered Cybersecurity: Emerging Trends and Challenges – A Narrative Review

View Full Paper

Ask AI

Bookmark

Authors

OJOras Nasef JasimThi Qar University NHNoor Abdulkaadhim HamadThi Qar University

Discussion

Member takes

Overview

Systematic review reveals hybrid artificial intelligence models consistently outperform standalone architectures in threat detection, highlighting critical trade-offs in computational cost and...

Key Points

To evaluate artificial intelligence applications in cybersecurity, focusing on intrusion detection, malware analysis, and the divergence between benchmark laboratory results and operational deployments.
Conducted a systematic literature review comparing standalone machine learning and deep learning algorithms against hybrid architectures (including CNN–LSTM and graph neural networks).
Evaluated models across accuracy, interpretability, computational efficiency, and real-world generalizability using existing benchmark datasets and evaluation protocols.
Hybrid, task-specific, and cross-task feature-learning architectures consistently surpassed standalone algorithms in threat detection accuracy, but incurred high computational costs and reduced model interpretability.
Significant divergence emerged between laboratory evaluations and operational cybersecurity environments, driven by over-reliance on static benchmarks, class-imbalanced data, vulnerability to adversarial attacks, and poor cross-domain transferability.

Cite This Study

Jasim et al. (2026) studied this question.

synapsesocial.com/papers/6aa7b2e10926e14a848b1676 https://doi.org/10.25195/ijci.v52i2.808

View Full Paper

Ask AI

Bookmark

September 14, 2026Iraqi Journal for Computers and InformaticsOpen Access

AI-Powered Cybersecurity: Emerging Trends and Challenges – A Narrative Review

View Full Paper

Ask AI

Bookmark

Authors

OJOras Nasef JasimThi Qar University NHNoor Abdulkaadhim HamadThi Qar University

Discussion

Member takes

Overview

Systematic review reveals hybrid artificial intelligence models consistently outperform standalone architectures in threat detection, highlighting critical trade-offs in computational cost and...

Key Points

To evaluate artificial intelligence applications in cybersecurity, focusing on intrusion detection, malware analysis, and the divergence between benchmark laboratory results and operational deployments.
Conducted a systematic literature review comparing standalone machine learning and deep learning algorithms against hybrid architectures (including CNN–LSTM and graph neural networks).
Evaluated models across accuracy, interpretability, computational efficiency, and real-world generalizability using existing benchmark datasets and evaluation protocols.
Hybrid, task-specific, and cross-task feature-learning architectures consistently surpassed standalone algorithms in threat detection accuracy, but incurred high computational costs and reduced model interpretability.
Significant divergence emerged between laboratory evaluations and operational cybersecurity environments, driven by over-reliance on static benchmarks, class-imbalanced data, vulnerability to adversarial attacks, and poor cross-domain transferability.

Cite This Study

Jasim et al. (2026) studied this question.

synapsesocial.com/papers/6aa7b2e10926e14a848b1676 https://doi.org/10.25195/ijci.v52i2.808

View Full Paper

Ask AI

Bookmark