Personal bias was previously considered rare in the law, but research has revealed that it is a significant issue within the legal system. Thanks to Natural Language Processing, now it is easier to systematically study and identify biases in case law, legislative materials, judgments, and sentencing decisions. The authors present how text analysis and natural language processing help find and analyze bias hidden within legal documents. By examining and reviewing recent peer-reviewed research published between 2020 and 2025, a critical evaluation of model frameworks, data used, and metrics to judge fairness has been performed. Transformer models, especially BERT, and fairness toolkits like AIF360 and FairLex have garnered significant attention. It was discovered that NLP technology can detect subtle bias in legal texts, but issues keep coming up when data, its use in other countries, and understanding its predictions are concerned. It has been shown that bias detection using NLP is possible and important, so extra research and implementation in the legal field are now needed for better justice transparency.