Which technique trains AI models across devices without sharing raw data?

• A. Federated learning
• B. Personal Information Protection Law (PIPL - China)
• C. GDPR (EU)
• D. EU AI Act of 2024

Answer: (A)

Federated learning enables training AI models across devices while keeping raw data on each device. In this approach, each device trains the model locally on its own data and only shares the updated model parameters (or gradients) with a central aggregator. The server combines these updates to form a new global model, which is then sent back to devices for another round. Because the actual data never leaves the device, this technique helps protect privacy while still enabling collaborative learning across many devices. Privacy can be further enhanced with secure aggregation or differential privacy, but the core idea is that raw data stays where it is while the model improves through coordinated updates.

The other options describe laws and regulatory frameworks about data privacy or AI governance rather than a method for training models. They govern how data can be used and how AI systems should operate, not the mechanics of distributed model training.