MeitY launches DEPA handbook for AI-training data sharing

MeitY’s handbook sets out a techno-legal framework for institutions to collaborate on AI-training data while retaining control of their data.[1] Its release addresses concerns that privacy, confidentiality, intellectual property and institutional control can limit data sharing.[1] Krishnan said India’s AI ambitions require data to be used responsibly and securely.[1] The Bengaluru roundtable drew representatives from government, industry, academia and research.[1] The iSPIRT Foundation supported development of the handbook.[1]

The framework combines contractual permissions, controlled computing environments and privacy-enhancing safeguards.[1] Its four technical pillars are a Data Exchange Platform, Electronic Contracts, a Secure Data Environment and Differential Privacy.[2] Under DEPA, the model travels to the data rather than the reverse.[1] A proposed plan also describes secure data environments, differential privacy and machine-readable electronic contracts, while keeping raw data inaccessible to developers.[2] The proposed sharing procedure says personal and sensitive data would not be shared in raw form.[2]

DEPA tracks data provenance so that a dataset’s origin, permissions and use can be followed throughout AI training.[1] The release says this tracking is intended to support accountability and transparency.[1] The framework proposes bias, security and explainability checks for trained models before deployment.[2] Together, these provisions describe controls around both the data used in training and the models produced; they do not, by themselves, establish how any particular institution will implement those controls.[1][2]

Under the proposed framework, providers such as hospitals, banks and public agencies would register datasets, while startups, researchers and enterprises would request access for specific use cases.[2] Intermediaries called Data Collabs would aggregate and standardise data from different sources.[2] The rollout roadmap starts with pilots in finance and healthcare alongside regulatory sandboxes.[2] Sood said DEPA enables multi-party and cross-border collaboration, and described it as India’s contribution to global science diplomacy.[1]