India unveiled a DEPA for AI Training handbook in Bengaluru on October 30, setting out a proposed framework for startups, researchers and companies to train models using institutional data with privacy safeguards. The approach aims to reduce the need to move sensitive raw data out of hospitals, banks and public agencies. It is a framework under development with pilots and further consultation, not a binding new law or a blanket right to access such data.
The initiative was developed with iSPIRT and presented at an event attended by MeitY Secretary S Krishnan and Principal Scientific Adviser Ajay Kumar Sood, according to Business Standard and Mint. DEPA for Training builds on work that predates this week's handbook: iSPIRT had already published technical updates and sample training scenarios earlier in 2026. The new event should not be mistaken for the invention of the underlying approach.
The proposed architecture brings together a data exchange, electronic contracts, secure data environments and differential privacy. Providers would make datasets available for defined use cases, while model builders would work within agreed controls. Business Standard describes more than one access arrangement, so the proposal should not be simplified into a claim that all data use follows one identical technical process or that privacy risk disappears.
Finance and healthcare are early focus areas. Mint reports that pilot demonstrations were shown at the event, while the roadmap includes further pilots and regulatory sandboxes. AI training also needs to be distinguished from inference: training builds or improves a model, whereas inference uses a model to produce a result. The framework's ambitions do not establish national deployment, commercial adoption or guaranteed outcomes.
FundForge Insight: for Indian AI builders, access to useful domain data can matter as much as access to compute. But a secure environment is only part of the answer. Data quality, enforceable contracts, liability when something fails and credible independent oversight will determine whether institutions and model builders can trust the system. Pilot evidence and clear rules matter more than large forecasts of economic value.
Sources: Business Standard (October 30, 2026): https://www.business-standard.com/technology/artificial-intelligence/india-proposes-framework-to-unlock-data-for-ai-training-protect-privacy-126103001392_1.html ; Mint (October 30, 2026): https://www.livemint.com/news/india-pilots-data-grid-to-train-ai-on-sensitive-info-without-seeing-it-11793362037802.html ; iSPIRT technical background: https://pn.ispirt.in/depa-training-tech-updates/ . Original FundForge typographic artwork; no third-party photograph. This is news analysis, not investment or legal advice.
