Investigating 5G Network Slicing Security Vulnerabilities Using Artificial Intelligence–Driven Intrusion Detection for Telecommunication Resilience

The implementation of network slicing in fifth-generation (5G) mobile networks enables the logical partitioning of physical infrastructure into multiple virtualized slices tailored for distinct service requirements such as enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communications (URLLC), and massive Machine-Type Communications (mMTC). However, this dynamic virtualization layer expands the system’s attack surface, introducing novel security vulnerabilities including slice isolation breaches, side-channel attacks, rogue slice instantiation, and service orchestration tampering. This review examines these vulnerabilities through a layered security perspective—spanning the radio access network (RAN), transport, and core domains—and analyzes how artificial intelligence (AI)-driven intrusion detection systems (IDS) can mitigate them. The study evaluates deep learning architectures such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Graph Neural Networks (GNN) for detecting anomalous inter-slice traffic and malicious orchestration behaviors within Software-Defined Networking (SDN) and Network Function Virtualization (NFV) environments. Moreover, the paper proposes a hybrid AI-IDS framework leveraging feature extraction from 5G control and user plane packets, unsupervised clustering for zero-day anomaly detection, and reinforcement-learning-based adaptive response. Experimental validation using the 5G-TONIC and Aalto University open datasets demonstrates over 96% detection accuracy with reduced false alarm rates under real-time conditions. The findings contribute to resilient 5G network orchestration and establish a foundation for adaptive threat intelligence in forthcoming 6G architectures.