In a significant move to shape its artificial intelligence ambitions, India has introduced a new framework for data collaboration. The guide aims to solve a critical puzzle: how to fuel AI innovation with data while protecting citizens' rights.
A New Rulebook for the AI Era
The Ministry
of Electronics and Information Technology (MeitY) has launched the 'Handbook on Data Empowerment and Protection Architecture (DEPA) for AI Training'. Unveiled at a roundtable in Bengaluru in late October 2026, the handbook lays out a techno-legal framework designed to build trust in the digital ecosystem. It addresses the persistent challenge of accessing diverse, high-quality data for training AI models, a process often stalled by concerns over privacy, intellectual property, and institutional control. The guidance is a critical piece of India's broader digital strategy, which includes the Digital Personal Data Protection Act (DPDPA) and the overarching IndiaAI Mission.
The Core Idea: Model to the Data
The central concept of the DEPA for AI Training framework is a paradigm shift from traditional data-sharing methods. Instead of raw data travelling to AI developers, the AI model travels to the data. As explained by Principal Scientific Adviser to the Government of India, Ajay Kumar Sood, this approach allows data-holding institutions like hospitals, banks, and government agencies to keep their sensitive datasets within a secure environment. AI developers, or 'Training Data Consumers', can then train their models on this data without directly accessing or moving it, thus preserving confidentiality and control. This method is designed to enable trusted, multi-party collaboration, even across borders, accelerating research and development.
Putting Data Rights First
The framework places a strong emphasis on data rights, privacy, and security. It proposes a combination of contractual permissions, controlled computing environments, and privacy-enhancing safeguards to ensure data is used responsibly. A key feature is the incorporation of 'data provenance', which allows the origin, permissions, and use of data to be tracked throughout the training process. This creates a clear audit trail, enhancing accountability and transparency. MeitY Secretary S. Krishnan noted that India's AI ambitions must be matched by the ability to use data securely, a principle at the heart of the new handbook.
Unlocking Value and Innovation
The government and industry experts believe this framework has the potential to unlock significant economic value. Projections suggest that enabling secure data access for AI could add up to $500 billion to India's GDP by 2030. For a long time, valuable datasets have remained locked in institutional silos, hindering innovation for Indian startups and researchers. By creating a trusted bridge, the DEPA framework aims to level the playing field, reduce market concentration favouring large entities, and spur the development of homegrown AI solutions in critical sectors like healthcare, finance, and agriculture. The initiative, developed with support from the iSPIRT Foundation, is seen as a foundational step in building a robust and ethical data economy.
