The spam filters, chatbots, and recommendation tools that have made artificial intelligence a fixture of modern life got there on data — mountains of training examples scraped from the web, or contributed by consumers in exchange for free email, music, and other perks.

Many of these AI applications were trained on data gathered and crunched in one place. But today’s AI is shifting toward a decentralized approach. New AI models are being trained collaboratively on the edge, on data that never leave your mobile phone, laptop, or private server.

This new form of AI training is called federated learning, and it’s becoming the standard for meeting a raft of new regulations for handling and storing private data. By processing data at their source, federated learning also offers a way to tap the raw data streaming from sensors on satellites, bridges, machines, and a growing number of smart devices at home and on our bodies.