Credit card fraud poses a significant challenge in financial transactions, necessitating the development of robust detection systems. This paper introduces an approach utilizing neural networks and adversarial training for credit card fraud detection. The proposed model leverages deep learning techniques to learn intricate patterns and detect fraudulent transactions effectively. By preprocessing the dataset, constructing a neural network model with appropriate layers, and training it using adversarial examples generated through perturbation, the model enhances its resilience to adversarial attacks. Experimental results demonstrate improved accuracy and robustness, contributing to secure transactions and preventing financial losses in the banking and financial sectors.