Enhancing Cybersecurity with AI: Predictive Threat Detection Using LLMs

Abstract

The rising frequency and complexity of cyber threats, including phishing, malware, and social engineering attacks, have made traditional cybersecurity methods are not enough. This paper explores how Artificial Intelligence (AI), especially predictive models and Large Language Models (LLMs), can improve threat detection and response. Predictive AI techniques, such as supervised and unsupervised learning, help identify anomalies and potential breaches by learning from historical and live network data. [3], [4], [6] At the same time, LLMs provide strong natural language understanding. This ability allows for detecting phishing attempts, analyzing complex malware scripts, and extracting threat intelligence from unstructured text sources.[8], [9]. This study gives a broad overview of AI-based cybersecurity methods, focusing on their real-world applications, effectiveness, and integration challenges. It also addresses limitations like Weak spots that attackers can trick, explainability, and ethical issues. The findings show the transformative potential of AI in creating flexible, smart, and scalable cyber defense systems