Federated learning is a potential solution for developing machine-learning models requirining massive datasets. However, it may have positive impacts on data protection such as avoiding data centralisation and simplifying control over data processing.
Learn more on federate learning with the EDPS here