Ransomware has become one of the most pervasive and damaging cyber threats, targeting individuals, enterprises, and critical infrastructures by encrypting essential data and demanding ransom payments. Traditional signature-based and heuristic detection methods are increasingly ineffective due to the rapid evolution, obfuscation techniques, and polymorphic behavior of modern ransomware variants. This research focuses on developing an advanced machine learning model for accurate, real-time, and adaptive ransomware detection. The study begins with a comprehensive review of existing ransomware detection approaches and their limitations. A hybrid detection framework is then designed using both static and dynamic features extracted from executable files and runtime behaviours. The model is implemented using appropriate artificial intelligence and machine learning algorithms to enhance detection accuracy and resilience. Experimental evaluation compares the proposed model with existing techniques, demonstrating improved performance in terms of accuracy, precision, recall, and robustness against zero-day ransomware attacks. The findings highlight the potential of advanced ML-driven approaches in strengthening cyber security defences and mitigating the growing impact of ransomware.