India proposes DEPA framework to train AI models on local data without moving it

MeitY's DEPA for AI Training framework proposes a 'model travels to the data' architecture using Training Data Providers, Data Exchange Platforms and Training Data Consumers, with hardware-based secure data environments, differential privacy and immutable audit logs. The source describes it as a policy proposal and does not give CVE identifiers, technical specifications or an implementation timeline. Defenders should treat it as a policy development to monitor rather than a current control.

For security analysts · Written by AI from 1 source · Wrong?