Högskolan i Skövde

This structured literature review (SLR) was conducted with the intention of investigating the practical challenges and opportunities associated with enhancing network security with the help of artificial intelligence (AI). The purpose of the study was to assess how AI-based approaches are being integrated today (2025) in enterprise networks, not only in theoretical models or in controlled laboratory environments, but in actual network environments with real-world conditions. Systematic searches were conducted in five different academic databases, for which a total of 983 articles were collected. These articles went through a screening process using predefined inclusion and exclusion criteria to objectively include or exclude the articles. After the screening process, 45 articles remained in which the key insights identified proceeded to be categorized into one of three main themes: key challenges for integration of AI into network security, the benefits AI can offer to network security, and real-world implementation cases. The results revealed that, while AI has the potential to significantly improve areas such as detection accuracy, response times, and efficiency in terms of operational performance. Adopting AI-based solutions tend to be hindered by issues related to explainability, data quality, scalability, and existing vulnerabilities to adversarial attacks. However, it was also revealed that AI-based models, especially those models using deep learning, reinforcement learning, and explainable AI techniques, are already being deployed today and show promising results in their current state. This study concludes that AI can be deployed today in a variety of different contexts, allowing for improvements in many areas of network security. However, for integrations to be successful, they must be done with careful planning, building trust in these systems, and by deploying these futuristic solutions gradually while not rushing the process.