Jamil, M.Creutzburg, R.2025-04-152025-04-15202597830318536232367-3370https://doi.org/10.1007/978-3-031-85363-0_24https://hdl.handle.net/20.500.12469/7284Securing critical infrastructures is essential to reducing risks in the rapidly evolving digital world. Traditional manual techniques of threat identification during cyberattacks are becoming less and less effective due to the limitations of human labor and the necessity for prompt responses. AI-based threat detection is a powerful solution that uses AI to identify, classify, and mitigate the effects of cyberattacks. Over the past five years, selecting appropriate AI and machine learning algorithms to evaluate threats in critical infrastructure protection has grown to be a significant challenge. Moreover, AI-driven threat detection must be seamlessly integrated into critical infrastructure cybersecurity. This work proposes a Supervised Learning model, a type of machine learning where the algorithm is trained on a labeled dataset, called the Random Forest algorithm for threat detection. The procedure entails thorough preprocessing and data accumulation from the NSL-KDD vulnerabilities database. The Random Forest model, known for its reliability, analyzes refined data and is skilled in identifying current risks and forecasting future ones. The study showcases the high accuracy and reliability of the model, with an accuracy score of 99.90% and a false positive rate of less than 15% for every assault category. These results underscore the effectiveness of the research in producing a reliable and accurate cybersecurity model. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.eninfo:eu-repo/semantics/closedAccessArtificial IntelligenceCritical InfrastructureCyber SecurityRandom ForestThreat DetectionConference Object3883981284 LNNS10.1007/978-3-031-85363-0_242-s2.0-105000878737N/AQ4