The classic centralized training of artificial intelligence (AI) models is experiencing privacy concerns as data are being housed in more and more silos and societies are becoming increasingly aware of data privacy issues. In addition to training robust global models that achieve exceptional accuracy levels, designing Machine Learning (ML) systems with privacy guarantees has become critical. With the advent of federated learning, collaborative model training across decentralized devices has become a powerful paradigm in the realm of machine learning. This presents numerous potentials in crucial fields where disclosing personal user information to other entities is risky and could not be done before. However, this distributed approach introduces a new avenue for potential privacy breaches, specifically through inference attacks where models can be intercepted and analyzed by adversaries both inside and outside the system, disclosing by that sensitive and/or personal information. In this research, we delve into the potential dangers posed by threat models, exploring the various methods through which malicious actors can intercept and exploit federated models to glean sensitive information. First, we underscore the significance of safeguarding federated learning systems against such threats, as compromised models may reveal sensitive patterns, jeopardizing user privacy and confidentiality. Then, the focal point of this study is the application of encryption techniques as a robust defense mechanism to fortify federated learning against inference attacks. By employing cryptographic protocols, we aim to elucidate how encryption can secure model parameters and predictions, preventing unauthorized access and maintaining the integrity of federated learning processes. Our research confirms that integrating encryption into federated learning has a manageable computational and time complexity, without sacrificing model accuracy. This subtle trade-off between security and performance not only strengthens data privacy but also bolsters overall security measures.