The extensive use of cloud computing for large-scale data processing has heightened concerns regarding data privacy and security. This paper examines the prime risks to data confidentiality in cloud-based big data environment which include identity disclosure, linkage attack and compliance risks. This study majorly focus on anonymization techniques as a primary mitigation strategy for privacy in cloud environment. In this paper, a detailed review of some widely used techniques, such as k-anonymity and l-diversity is done. The experimental evaluations is presented using real-world datasets, such as UCI Adult dataset by processing in a distributed Hadoop environment. Key performance indicators include privacy risk, data utility and processing overhead that is measured under different anonymization strategies and configurations. The empirical findings clearly present the trade-off between data utility and privacy preservation, showing that by using adaptive tuning and hybrid optimization approaches, it is possible to significantly enhance protection with little loss in analytical value. The work closes by referring to open challenges and by calling for intelligent, metaheuristic-assisted anonymization frameworks for handling emerging privacy demands of scalable cloud infrastructures. Finally, potential solutions and future research directions are discussed in order to improve big data privacy and security in cloud environment.