Detection of browser attacks is considered a serious challenge in today’s web applications. Man in the Browser (MitB) attack is an important type of these attacks that can lead to changes in web page contents, interference in network traffic, session hijacking, and user information theft by using Trojans. In this paper, an efficient tool for real-time detection of MitB attacks through dynamic analysis of web pages based on the description of attack patterns is presented. The advantage of the proposed tool is that it is not limited to identifying one or more specific attacks and the identification method code is not embedded in the tool, but the patterns of different attacks are specified separately. In order to evaluate the presented tool, two vulnerable web services provided by OWASP, which have a wide range of known vulnerabilities, were used along with the BeEF penetration test framework, and a set of MitB attacks were practically implemented and evaluated by the tool. The same tests were performed using three other similar tools and compared with the developed tool. In addition to the superiority of the presented tool in terms of the independence of attack descriptions from the tool itself, the results show that the accuracy and readability of its diagnosis are better than similar tools.

As the intensity of global cybersecurity threats continues to rise, the need for training security professionals has gained greater significance. Educational programs, complemented by laboratories and the execution of cybersecurity exercises, play a fundamental role in enhancing both offensive and defensive capabilities. The execution of such exercises is particularly crucial in operational networks, where testing cyberattacks may not be feasible. Cyber ranges offer an appropriate platform for conducting these exercises. A primary challenge in cybersecurity education is aligning training programs with the diverse skill levels of learners. Adaptive learning, powered by artificial intelligence and recommendation systems, can provide an effective solution for delivering personalized instruction. This study focuses on the KYPO Cyber Range to examine the potential of substituting or augmenting the role of the instructor with an AI-based recommendation agent. The objective of this research is to minimize human intervention and improve the efficiency of the training process. To this end, data collected from the KYPO Cyber Range, developed by Masaryk University, has been utilized, and various machine learning models have been applied to automate and optimize the training process. The results of this research indicate that the integration of artificial intelligence can enhance the performance of educational systems and reduce evaluation time.