The rapid expansion of digital networks and internet-connected devices has significantly increased the exposure of organizations to sophisticated cyber-attacks. Traditional signature-based Intrusion Detection Systems (IDS) struggle to identify novel and evolving threats due to their reliance on predefined attack patterns. This paper proposes an AI-based Network Intrusion Detection System (NIDS) that leverages machine learning algorithms to detect malicious network traffic with high accuracy and low false-positive rates. The proposed system employs data preprocessing, feature selection, and classification techniques including Decision Tree, Random Forest, Support Vector Machine, Naïve Bayes, and Artificial Neural Networks, trained and evaluated on benchmark intrusion detection datasets. Experimental results demonstrate that ensemble-based approaches, particularly Random Forest, achieve superior detection accuracy compared to conventional methods. The study concludes that integrating machine learning into intrusion detection significantly enhances the ability of networks to identify both known and previously unseen attacks in real time.

Keywords— Network Intrusion Detection System, Machine Learning, Cybersecurity, Random Forest, Support Vector Machine, Anomaly Detection, Network Security.