AI for Beyond 5G Networks: A Cyber-Security Defense or Offense Enabler?

Artificial intelligence (AI) is envisioned to play a pivotal role in empowering intelligent, adaptive and autonomous security management in 5G and beyond networks, thanks to its potential to uncover hidden patterns from a large set of time-varying multi-dimensional data, and deliver faster and accurate decisions. Unfortunately, AI's capabilities and vulnerabilities make it a double-edged sword that may jeopardize the security of future networks. This article sheds light on how<;span class="upper-case"> Ai <;/span>may impact the security of 5G and its successor from its posture of defender, offender or victim, and recommends potential defenses to safeguard from malevolent AI while pointing out their limitations and adoption challenges.

Introduction

5G and beyond networks hold out the promise of delivering ultra-low latency, ultra-high throughput, ultra-high reliability, ultra-low energy usage, and massive connectivity. Achieving these promises will pave the way to a new breed of applications, including autonomous driving, industry 4.0, augmented and virtual reality, collaborative gaming, near real-time remote surgery, and teleportation. However, the diversity of services/applications and the growing number of connected things envisaged in the networks of tomorrow will open up new and increasingly broad cyber threats, posing security and privacy risks [1]. Thus, it is imperative to build up effective and sustainable security measures that can deal with the ever-evolving threat landscape and security requirements in 5G and its successor in order to fully reap their benefits.