In a major push for its artificial intelligence ambitions, India has unveiled a new handbook designed to solve one of the most critical challenges in the AI race: how to share vast amounts of data for innovation while keeping it secure and private.
A New Rulebook for the Data Economy
The
Ministry of Electronics and Information Technology (MeitY) has launched the 'Handbook on Data Empowerment and Protection Architecture (DEPA) for AI Training'. Unveiled in Bengaluru, this initiative provides a techno-legal framework that allows organisations like hospitals, banks, and public agencies to collaborate by making their data available for training AI models. The goal is to break the 'data deadlock' where valuable information remains locked within institutional silos due to fears over privacy, intellectual property, and commercial confidentiality. Developed with support from the iSPIRT Foundation, the handbook is not a binding law but a set of guiding principles to foster a trusted environment for data sharing between 'Training Data Providers' and 'Training Data Consumers' like startups and researchers.
Why Was This Framework Needed?
Access to diverse, high-quality data is the fuel for a powerful AI ecosystem. However, Indian organisations have faced significant hurdles. Studies show that while most companies have AI data policies, breaches are common, and a lack of data standardisation is a major barrier for AI developers. The DEPA for AI Training framework addresses this by creating a structured approach to data sharing that moves beyond simple anonymisation, which is often insufficient to prevent re-identification. The handbook acknowledges that for innovation to flourish, there must be a clear, trusted mechanism for collaboration. The government has stated that inaction is not a neutral stance; without a framework, India risks stifling innovation for startups and increasing market concentration among large entities that already hold massive datasets.
How It Works: 'Model Travels to the Data'
A core principle of the DEPA framework is a paradigm shift in how data is accessed. As Principal Scientific Adviser Prof. Ajay Kumar Sood explained, the goal is for 'the model to travel to the data, rather than the data travelling to the model'. This means AI developers won't get their hands on the raw data. Instead, the framework proposes the use of 'secure data environments' where AI models can be trained without exposing sensitive information. This is combined with other privacy-enhancing techniques like differential privacy, which adds statistical noise to data to protect individual identities. The framework also incorporates 'data provenance', a system to track the origin, permissions, and use of data throughout the AI training process, ensuring transparency and accountability.
Unlocking Billions in Economic Value
The ambitions tied to this handbook are immense. The document estimates that responsibly unlocking data for AI could add approximately $500 billion in value to India's GDP by 2030. This economic potential extends across numerous sectors. In finance, for example, better access to broader datasets could help financial institutions more accurately assess creditworthiness, potentially extending formal credit to millions of underserved individuals and small businesses. By lowering the barrier to data access, the framework aims to level the playing field, allowing smaller startups to compete with tech giants and drive innovation in fields like healthcare, agriculture, and public services. This initiative complements India's broader push to build a comprehensive digital public infrastructure for the AI era.
A Stepping Stone in AI Governance
The DEPA handbook is a crucial component of India's larger strategy for AI governance. It arrives as the country navigates the complexities of its Digital Personal Data Protection (DPDP) Act, which some experts note does not fully address unique AI challenges like algorithmic bias and transparency. Industry bodies like NASSCOM have advocated for a collaborative and evidence-based approach to AI regulation, favouring coordination and voluntary frameworks over rigid, topdown laws. The DEPA handbook aligns with this philosophy, providing a practical blueprint for responsible behaviour that can evolve with the technology. It represents India's effort to forge a path that balances rapid innovation with the paramount needs of safety, trust, and privacy, positioning itself as a leader in the global conversation on responsible AI.
