Federated Learning (FL) is a machine learning approach that enables multiple parties to collaboratively train a shared model using their local data without sharing the data itself. Model training occurs on a set of client devices, such as mobile phones, while maintaining the data locally. Thus, the data does not leave the client devices, and its privacy is naturally protected. Only the model updates, which are often of much smaller sizes than the data itself, in addition to other information, such as model architecture and initialization parameters, get transferred back and forth between the parties. Such a design can significantly reduce the communication costs. Federated Learning also takes advantage of the data distribution of client devices to obtain better performance by training on data at a larger scale while augmenting new types of data, as new clients can continuously join in new training rounds. As Federated Learning continues to offer new possibilities for privacy-preserving and cooperative settings, current privacy protection mechanisms may not be compatible with Federated Learning. Notably, while Federated Learning presents a privacy-preserving way of working with user data, it does demand access to user data for computation and personalization. Despite data remaining on user devices, it still runs on user data directly, making it possibly subject to regulations for medical data, controls placed on European Union residents regarding all data, and for US residents, among numerous other global privacy regulations. With globally deployed Federated Learning systems, a clearer understanding and expanded regulation collaboration can help alleviate compliance concerns and make Federated Learning a standardized process.