MeitY Launches DEPA Handbook in Bengaluru to Facilitate Secure Data Sharing for AI Training

The Union Ministry of Electronics and Information Technology (MeitY) has introduced a new framework designed to facilitate secure data collaboration for artificial intelligence development while addressing concerns related to privacy, intellectual property, and institutional control.

The initiative was unveiled on Saturday at a high-level roundtable in Bengaluru by MeitY Secretary S. Krishnan, with support from the iSPIRT Foundation. The newly launched Handbook on Data Empowerment and Protection Architecture (DEPA) for AI Training establishes a techno-legal model for purpose-bound data sharing, allowing institutions to participate in AI training processes without losing authority over their data.

The framework integrates contractual permissions, controlled computing environments, and privacy safeguards. It also features data provenance tracking to monitor the origin, permissions, and usage of datasets throughout the AI training lifecycle, ensuring transparency and accountability.

Speaking at the launch, S. Krishnan stated that the framework is designed to match India's artificial intelligence goals with responsible data practices, enabling trusted collaboration at scale and unlocking access to large datasets while safeguarding commercial confidentiality and privacy.

Ajay Kumar Sood, Principal Scientific Adviser to the Government of India, noted that the architecture allows AI models to access information securely without necessitating the transfer of underlying datasets. Under this approach, the model travels to the data rather than the data moving to the model, which could support multi-party and cross-border scientific collaboration.

Sharad Sharma, Co-founder of the iSPIRT Foundation, compared the significance of the framework to the introduction of the Unified Payment Interface (UPI) in February 2016. Additionally, Kris Gopalakrishnan, Co-founder of Infosys and Chairman of Axilor Ventures, remarked that the framework could assist researchers in accessing diverse datasets for robust AI model development while securing the rights of data custodians.

The roundtable event convened stakeholders from government, industry, technology, academia, research, and media sectors to discuss the application of trusted data collaboration in advancing innovation and public goods.

"The introduction of the DEPA framework by MeitY addresses a fundamental friction point in technology development: how to scale AI training while preserving data ownership, security, and privacy. By establishing a techno-legal model where models access data rather than transferring underlying assets, this initiative provides a structured foundation for institutional data collaboration. For the broader tech and startup ecosystem, clear data provenance and controlled computing environments can significantly reduce compliance risks and encourage wider cross-sector partnerships in artificial intelligence innovation." — Dr. Shishir Gupta, Founder & CEO, StartupLanes