The increasing complexity of financial transactions driven by global economic integration and advancements in technology has made fraudulent activities increasingly challenging to detect. This paper presents an innovative approach leveraging Graph Convolutional Networks (GCNs) to detect financial fraud within the Elliptic Bitcoin Dataset. By combining distributed data preprocessing, advanced graph construction, and feature engineering techniques, our method captures complex relational patterns and temporal dynamics inherent in transaction networks. The proposed solution leverages GCNs to enhance the detection of fraudulent transactions while addressing challenges such as data imbalance and feature weakening through adaptive filtering mechanisms. Experimental validation demonstrates the robustness of the proposed model, achieving an accuracy of 95% on the dataset, while highlighting challenges related to minority class detection. This paper discusses the implications of these findings and proposes future research directions to improve fraud detection systems.