India's MeitY Releases Guide to Train AI Without Sharing Data

India's Ministry of Electronics and Information Technology (MeitY) has unveiled a handbook that explains how to train artificial intelligence models without sharing the underlying data.

The guide was developed with support from the iSPIRT Foundation, an Indian tech think tank. It addresses a problem that many organisations face when building AI: the data needed to train useful models is often sensitive, regulated or held by different parties who are reluctant to hand it over.

Why it matters

By setting out approaches that avoid pooling or transferring raw data, the handbook gives companies, startups and public bodies in India a reference point for AI projects that involve confidential information. That is particularly relevant in sectors where records are personal or commercially valuable.

The publication also signals that the government is looking at practical tools for AI development alongside policy, with a think tank involved in shaping the guidance.

Further details of the handbook's contents and how it will be applied were not available at the time of writing.