The Ministry of Electronics and Information Technology (MeitY) has released a crucial new guide for India's booming AI sector. This 'techno-legal' handbook aims to solve a major hurdle: how to train powerful AI models using sensitive data without compromising
privacy or intellectual property.
A Rulebook for the AI Revolution
On October 31, MeitY unveiled the 'Handbook on Data Empowerment and Protection Architecture (DEPA) for AI Training'. Announced in Bengaluru by IT Secretary S Krishnan, the framework was developed with support from the iSPIRT Foundation. It's not a new law, but a detailed guide that creates a structured way for institutions like banks, hospitals, and government agencies to collaborate on AI development. The core problem it addresses is that while high-quality data is essential for building effective AI, concerns over privacy, confidentiality, and commercial secrets often prevent this data from being shared. This handbook provides a framework to unlock these valuable datasets, fuelling India’s AI ambitions while keeping data secure.
How the Framework Operates
The DEPA for AI Training introduces a novel concept: instead of data travelling to the AI model, 'the model travels to the data'. This means AI developers won't get to move or copy the raw datasets. Instead, their models will be trained inside secure, controlled computing environments managed by the data-holding institutions. The framework combines several layers of protection. First, it relies on strong contractual agreements that limit data use to specific, agreed-upon purposes. Second, it uses privacy-enhancing technologies to protect sensitive information. And third, it incorporates 'data provenance', a system that tracks the origin, permissions, and usage of data throughout the training process, ensuring transparency and accountability.
Balancing Innovation and Safeguards
India's approach is to foster innovation through what it calls a 'light-touch' regulatory model. Rather than introducing a rigid, all-encompassing AI law, the government is building on existing legal frameworks. This includes the Digital Personal Data Protection Act, 2023, which governs the use of personal data, and the IT Rules, which address issues like deepfakes and synthetic media. The DEPA handbook is a key part of this strategy. IT Secretary S Krishnan noted that India's AI ambitions 'must be matched by the ability to use data responsibly and securely'. The framework aims to strike that balance, allowing startups and researchers to innovate while giving data providers the confidence that their information is protected.
What This Means for Startups and Industry
For India's burgeoning AI startup ecosystem, this is a significant development. Access to large, high-quality datasets is often a major barrier, especially for smaller companies that can't afford to purchase data. The DEPA framework could level the playing field, allowing startups to train sophisticated models for sectors like healthcare and finance using data that was previously inaccessible. At the launch event, representatives from major industry players like HDFC Bank and Infosys co-founder Kris Gopalakrishnan expressed support, signalling industrywide interest. By creating a trusted mechanism for data collaboration, the handbook has the potential to accelerate the development of Indiaspecific AI solutions in critical fields.
