A Survey on Real-Time Log Anomaly Detection Using Artificial Intelligence: Methods, Challenges, and Transparency Considerations
Year: 2026
Type: article
Abstract: Cybersecurity faces constant innovations by attackers, with increasingly intelligent and intricate techniques. In response, cybersecurity researchers have been exploring AI-based approaches to improve cyber defense systems. AI techniques are being applied to analyze vast amounts of log data, identifying patterns and anomalies that could indicate malicious activity. These AI-driven systems are crucial in improving the efficiency and effectiveness of threat detection. The purpose of this scoping review is to study how Artificial Intelligence (AI) is applied in the literature for the detection of real-time threats in cybersecurity logs. To this end, the methodology defined by Arskey and O’Malley for scoping reviews has been followed. A taxonomy has also been designed herein to methodically classify and organize the reviewed articles. The results obtained include the utilities used, information on explainability, the metrics employed, and a list of the most common errors in Machine Learning (ML) applications. In conclusion, despite the great advances in the use of AI in threat detection through log analysis, there remain certain areas that require extra attention, such as transparency in ML models through the integration of Explainable AI (XAI), standardization in log formats and greater availability of data.
Source: Electronics
Institution: Universidad de Sevilla
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,896
OpenAlex ID: W7234049344
