Mitigating Ransomware Threats: A Machine Learning Perspective on Dynamic Feature Utilization

In a review of the literature on detecting and averting ransomware attacks, this study focuses on current studies that were published between 2020 and 2025. Although the quick development of digital technology has brought about many conveniences, the threat of ransomware—a type of malware that encrypts the victim's files and then demands a large ransom to unlock them—is also growing. Cybersecurity is generally defined as safeguarding systems against all cyberattacks. Here, we know that ransomware is a method of stealing money from a user, where the attacker encrypts the user's data and retains the decryption key until he is paid the ransom. In this study, we examine a number of scholarly works that include ransomware detection strategies, indicators, tactics, approaches based on URL characteristics, and efficient machine learning models. This study takes into account new patterns in the accuracy of models from URL datasets that come from different machine-learning techniques as well as novel methods that use diverse models to identify ransomware. The results highlight the limitations of existing research and the application of robust models to develop cybersecurity hybrid models. Future research aims to choose appropriate models for identifying URL-based ransomware on Android smartphones based on this review. Even if there are still fresh, evident problems and restrictions with different detection methods, this article also emphasizes the necessity of constant development in the constantly evolving URL-based ransomware detection strategies.