The problem of recognizing credit card fraud is currently relevant due to the significant increase in the use of credit cards by the population. At the same time, the methods and algorithms used by credit card companies are far from perfect. Machine learning methods and algorithms are currently being used to solve this problem. The paper presents some current research being conducted in this area. This paper presents a study on the use of a neural network for credit card fraud detection. The availability of publicly available training datasets and the challenges of configuring a neural network based on organizational policies are discussed. It demonstrates ways to tune the neural network under consideration to better recognize fraudulent transactions as such, while observing a greater number of legitimate transactions classified as fraudulent, and vice versa. It also demonstrates ways to tune the neural network to minimize the classification of legitimate transactions as fraudulent, while observing the omission of fraudulent transactions.