In last one and half decade, Android has evolved as the most widely accepted and technologically resilient platform. However, as its popularity grew, the platform also became the target of malicious activities. Smart devices often contain users’ sensitive and private data, making them attractive targets for hackers. The growing attempts have targeted device integrity, financial protection, and file protection mechanism. To prevent this harm caused by malware, various techniques have been introduced e.g., signature-based, heuristic-based, and machine learning model-based, which are getting their models trained on centralized data. This centralized user data serves as a focal point for data leakage, raising significant concerns about users’ privacy. Confronting these challenges, the paper introduces a modern approach for android malware detection that harnesses the power of Federated Learning using machine learning model to preserve privacy. This study employs the XGBoost model after comprehensive preprocessing of the CICMalDroid2020 dataset, integrating federated learning across four client applications to ensure high performance while preserving user data privacy. Without transmitting data to a centralized server, the proposed federated framework demonstrates superior results compared to existing state-of-the-art approaches, achieving an impressive accuracy of 98.80% and higher F1-score.