The recent trends of cloud computing in businesses deliver various platforms, software, and services via the internet. Secure Authentication and Authorization methods must be implemented in cloud secure systems and networks to ensure that sensitive data and resources are only accessible to approved users. Therefore, this study aims to proactively identify and mitigate anomalies in authorization and authentication processes in cloud systems. This paper addresses the need for an enhanced anomaly detection system to strengthen the cloud security inside SAML and OAuth 2.0 frameworks. A real time anomaly detection system is put forward, which continuously monitors authentication and authorization activities patterns, analyzing them for deviations from established norms. To achieve this, unsupervised machine learning algorithms are used including, Isolation Forest, the system identifies anomalies indicative of potential security breaches and unauthorized access attempts, which enable cloud-based systems to be more resilient to possible security breaches. In addition, the strength of the proposed model is validated based on a real labeled data set and conduct a thorough evaluation of vulnerabilities unique to SAML and OAuth 2.0. Compared to existing techniques, the proposed solution exhibits a promised performance in terms of anomaly detection.