MeitY Launches DEPA Handbook to Enable Secure Data Sharing for AI Training

MeitY has introduced a DEPA handbook outlining a secure framework for institutions to share data for AI training while protecting privacy, intellectual property and institutional control.

MeitY Launches DEPA Handbook to Enable Secure Data Sharing for AI Training

The Ministry of Electronics and Information Technology (MeitY) has introduced a handbook on the Data Empowerment and Protection Architecture (DEPA), aimed at enabling secure and responsible data sharing for artificial intelligence training.

The handbook presents a techno-legal framework that allows organisations and institutions to collaborate using data without giving up control over their information. It seeks to address concerns surrounding privacy, confidentiality, intellectual property and data ownership that often restrict data-sharing arrangements.

IT Secretary S Krishnan unveiled the handbook at a roundtable in Bengaluru. The initiative was developed with support from the iSPIRT Foundation.

What Is the DEPA Handbook for AI Training?

The DEPA handbook provides guidelines for institutions that want to make their data available for AI training while maintaining appropriate safeguards.

As artificial intelligence becomes increasingly important across economic and social sectors, access to diverse and reliable datasets is essential for developing effective AI systems. However, organisations may hesitate to share information because of concerns about how it could be accessed, processed or used.

DEPA aims to address these challenges by establishing a structured approach to data collaboration. Its framework combines legal agreements with technical controls to ensure that data is used only under authorised conditions.

Key Features of the DEPA Framework

1. Contractual Permissions and Controlled Computing

The framework brings together contractual permissions and controlled computing environments to regulate how participating institutions can use shared data.

These arrangements are intended to ensure that organisations retain control over their information while allowing approved parties to collaborate on AI training activities.

2. Privacy-Enhancing Safeguards

DEPA incorporates privacy-focused protections to reduce risks associated with sharing sensitive or confidential information.

These safeguards are designed to support collaboration without compromising the privacy requirements and institutional responsibilities of participating organisations.

3. Data Provenance and Transparency

Another important component of the framework is data provenance. It enables the origin of data, the permissions attached to it and its usage throughout the AI training process to be tracked.

This approach can help establish accountability by making it easier to understand where training data came from and whether it was used according to the agreed conditions.

MeitY Says Secure Data Access Is Important for India's AI Ambitions

Speaking at the launch, IT Secretary S Krishnan highlighted the importance of responsible data usage in India's artificial intelligence development plans.

He said the country's AI ambitions must be supported by the ability to use data securely and responsibly. Krishnan also reaffirmed MeitY's commitment to advancing the DEPA framework.

The handbook is intended to help institutions overcome barriers to data collaboration while ensuring that privacy, confidentiality and control remain important considerations.

Model Travels to the Data Under DEPA Framework

Principal Scientific Adviser Ajay Kumar Sood explained a key principle behind DEPA: instead of transferring data to the model, the model travels to the data.

This approach allows AI models to work within controlled environments where the data is held, rather than requiring institutions to move their datasets to external locations.

According to Sood, the framework can support collaboration involving multiple organisations and even participants across national borders. He also described DEPA as India's contribution to global science diplomacy.

Bengaluru Roundtable Brings Together Key Stakeholders

The handbook was unveiled during a roundtable discussion in Bengaluru attended by representatives from government, industry, academia and research organisations.

The gathering provided an opportunity for stakeholders from different sectors to discuss the framework and its potential role in enabling secure data collaboration for AI development.

The involvement of these groups reflects the importance of cooperation between public institutions, businesses and research communities in building responsible AI capabilities.

Conclusion

MeitY's DEPA handbook aims to address one of the major challenges in AI development: enabling access to useful data without compromising privacy or institutional control. By combining contractual permissions, controlled computing environments and data provenance, the framework offers a structured approach to secure collaboration. Its adoption could support research partnerships and AI training initiatives in India while strengthening the country's focus on responsible data use.