What is the New Framework?
On October 31, 2026, the Ministry of Electronics and Information Technology (MeitY) launched the 'Handbook on Data Empowerment and Protection Architecture (DEPA) for AI Training'. Unveiled in Bengaluru, the initiative was developed with support from the iSPIRT
Foundation, a non-profit technology think tank. The handbook is not a new law but a techno-legal framework. It provides a structured approach for organisations to collaborate on AI projects by sharing data in a controlled and secure manner, aiming to protect privacy, intellectual property, and institutional control.
The Core Problem: A Crisis of Trust
High-quality data is the lifeblood of powerful AI systems. However, collaboration between companies, researchers, and startups is often stalled by significant concerns. Businesses are reluctant to share proprietary data for fear of losing their competitive edge or compromising customer privacy. A recent study noted that data security and accessibility are among the biggest obstacles to AI adoption in India. This handbook directly addresses this trust deficit, creating a pathway for 'Training Data Providers' (like hospitals and banks) to work with 'Training Data Consumers' (like startups and researchers) safely.
How It Works: 'Model to the Data'
A key concept in the DEPA framework is enabling 'the model to travel to the data, rather than the data travelling to the model'. This approach minimises the risks of data leakage. Instead of data holders sending their raw datasets to AI developers, the developers' AI models are brought into a secure, controlled computing environment where the data resides. The framework combines contractual agreements, privacy-enhancing technologies, and robust data provenance mechanisms. These mechanisms track the origin, permissions, and use of data throughout the AI training process, ensuring transparency and accountability.
Beyond Individual Consent
One of the handbook's most debated aspects is its approach to consent for large-scale AI training. It notes that seeking consent from every individual whose data is part of a massive dataset is often 'impractical'. To protect the interests of individuals, the framework relies on strong governance structures and technical safeguards rather than direct interaction. Intermediaries called 'Data Collabs' will be responsible for pooling data and obtaining primary consent where applicable. These collabs will be commercial entities, and the framework even suggests royalty or access-fee models to incentivise data providers.
Unlocking India's AI Potential
The DEPA for AI Training is a foundational piece of India's larger strategy to become a global leader in artificial intelligence. By creating a trusted ecosystem for data sharing, the government hopes to unlock innovation across critical sectors. For example, the handbook proposes combining health data from government portals to build disease outbreak warning systems or using de-identified financial data to extend credit to small businesses without formal loan histories. For MeitY Secretary S. Krishnan, the framework is essential for matching India's AI ambitions with the ability to use data responsibly and securely. The move is seen by some as a milestone comparable to the launch of the Unified Payments Interface (UPI), which revolutionised digital payments in the country.
