| Title |
Leveraging generative AI models for multidomain network security |
| Authors |
Brūzgienė, Rasa ; Grigaliūnas, Šarūnas ; Veitaitė, Ilona ; Danielienė, Renata ; Driaunys, Kęstutis ; Astromskis, Paulius ; Nemickienė, Živilė ; Vengalienė, Dovilė ; Stankūnas, Rokas ; Andrijauskaitė, Ieva ; Gaubienė, Neringa |
| DOI |
10.5220/0000216400004052 |
| ISBN |
9789897587962 |
| Full Text |
|
| Is Part of |
ICAART 2026: Proceedings of the 18th International Conference on Agents and Artificial Intelligence, Marbella, Spain. Vol. 5.. Setúbal : SciTePress - Science and Technology Publications, 2026. p. 3909-3916.. ISSN 2184-433X. ISBN 9789897587962 |
| Keywords [eng] |
GenAI ; network security ; TechSec ; OpSec |
| Abstract [eng] |
The rise of Generative Artificial Intelligence (GenAI) presents new opportunities for enhancing network security across technical, operational, human, and physical domains. This paper proposes a GenAI-driven network security framework that integrates OSI-layer–aware threat analysis with mandatory human oversight to support CISO decision-making. The framework is empirically evaluated using NetFlow datasets representing port scanning, ICMP flooding, and SPAM attacks. Six GenAI models are assessed using explainability index, hallucination rate, and token efficiency metrics. The results demonstrate that while certain models achieve high analytical accuracy and explainability, performance varies significantly in efficiency and hallucination behavior. The study further discusses legal and regulatory implications of deploying GenAI in security-critical environments, highlighting the necessity of human-in-the-loop control for accountable and reliable network security operations. |
| Published |
Setúbal : SciTePress - Science and Technology Publications, 2026 |
| Type |
Conference paper |
| Language |
English |
| Publication date |
2026 |
| CC license |
|
