Information Technology: Faculty Publications

AI-Based Detection of Zero-Day Exploits: A Framework

Document Type

Conference Proceeding

Publication Date

12-31-2025

Publication Title

Proceedings of the 2025 IEEE 12th International Conference on Intelligent Computing and Information Systems, ICICIS 2025

DOI

10.1109/ICICIS66182.2025.11313144

Abstract

Zero-day exploits remain one of the most pressing cybersecurity challenges, as they exploit software vulnerabilities that are unknown to developers and security teams, leaving systems vulnerable until a fix is released. This research proposes an AI-powered model for potential real-time detection of zeroday exploits in web browsers, which are major cybersecurity threats due to their ability to exploit unknown vulnerabilities with unknown signatures. The model uses machine learning, anomaly detection, and behavioral analysis framework to identify suspicious activity in real time. It continuously monitors browser behavior and system logs, detecting previously unseen threats without relying on prior knowledge. The model includes intelligent risk assessment using a five-level threat classification system and enables automated incident response. Our framework leverages the advanced detection capabilities of generative AI models, including OpenAI and Gemini-based algorithms to improve detection accuracy and substantially minimize false positives. This solution offers practical benefits for users and organizations by narrowing the vulnerability window between exploit discovery and mitigation. It also sets a foundation for future developments like self-healing browsers and decentralized AI security networks.

Comments

Georgia Southern University faculty member, Atef Shalan co-authored, "AI-Based Detection of Zero-Day Exploits: A Framework."

Link to Full Text

Share

COinS