Federated learning is a machine learning approach in which models are trained across separate devices or organizations without centralizing all raw training data.
It can support privacy-sensitive collaboration, though teams still need controls for security, leakage, bias, and model quality.