India Proposes Data Framework to Accelerate AI Model Development

India’s Ministry of Electronics and Information Technology (MeitY) has introduced a proposed framework designed to facilitate secure, consent-based data sharing for artificial intelligence model training. Developed alongside the think tank iSPIRT, the initiative aims to provide startups, businesses, and academic institutions access to large-scale, India-specific datasets while maintaining user privacy and compliance with the Digital Personal Data Protection (DPDP) Act.

Establishing the Architecture

The proposed Data Empowerment and Protection Architecture (DEPA) for AI training builds upon India’s existing consent-based data sharing infrastructure. The architecture employs technical pillars, including confidential clean rooms, which allow organizations to collaborate on training without exposing raw datasets. It also incorporates differential privacy to prevent individual information extraction, electronic contracts for automated enforcement of usage rules, and zero-trust security protocols.

Defining Stakeholder Roles

The framework categorizes participants into three specific roles. Training Data Providers, such as banks or hospitals, make their data accessible. Training Data Consumers, including researchers and enterprises, request this data for model development. Data Collectives act as intermediaries to aggregate, standardize, and govern the data exchange policies.

Operational Workflow

The process begins with providers registering anonymized and encrypted datasets on an exchange platform. These datasets are tagged with metadata and cryptographic identifiers. When a consumer requests data, an electronic contract system verifies the request against established governance rules. Access is granted within a secure environment, allowing model training to proceed without the raw data ever transferring to the consumer.

Current Status

The ministry is now advancing this techno-legal framework to enable organizations to collaborate on AI development without direct data transfers. The guidelines aim to foster innovation by creating a structured, safe pathway for researchers and businesses to utilize India-specific datasets.