MeitY unveils DEPA handbook to enable trusted data sharing for AI training

The Ministry of Electronics and Information Technology (MeitY) has unveiled a handbook on Data Empowerment and Protection Architecture (DEPA) aimed at enabling trusted data sharing for AI training, an official statement announced on Saturday. This handbook sets forth a techno-legal framework that empowers institutions to collaborate on data for AI training, while retaining control over their data.

The framework, developed with support from the iSPIRT Foundation, comprises contractual permissions, controlled computing environments, and privacy-enhancing safeguards. It also integrates data provenance, which allows the origin, permissions, and usage of data to be traced throughout the training process, thus promoting accountability and transparency.

With AI becoming pivotal to economic and social progress, access to a wide array of high-quality data is crucial. However, concerns over privacy, confidentiality, intellectual property, and institutional control frequently hinder data sharing. The DEPA handbook seeks to address these challenges. Speaking at the launch event in Bengaluru, IT Secretary S Krishnan highlighted the handbook's significance, emphasizing that India's AI ambitions must be accompanied by the capability to utilize data responsibly and securely. MeitY is committed to further advancing the framework.

Principal Scientific Adviser Ajay Kumar Sood explained that under DEPA, the AI model is trained on the data itself, rather than the model being moved to where the data resides. This approach facilitates multi-party and cross-border collaborations. Sood described DEPA as India's contribution to global science diplomacy. The roundtable discussion to launch the handbook included participants from government, industry, academia, and research institutions, underscoring its collaborative nature.