Large Language Models have presented an impressive performance and have become fundamental in real-world applications. Nevertheless, these models also exhibit certain limitations. They tend to absorb and potentially magnify social biases inherent in their training data. As a result, this may lead to biased outcomes in their applications, potentially resulting in negative social consequences. This chapter aims to present an in-depth literature of existing studies on biases and fairness in LLMs. It explores various existing studies and blogs to acquire the related content of biases and fairness. The primary contribution of the proposed chapter is to cover biases and fairness-specific metrics, benchmarks, datasets and strategies of biases mitigation. This chapter can provide a base to find the existing literature to the community on biases and fairness in LLMs.