AI Safety Institute
Also called: AISI · Topic: Statutory, Regulatory and Quasi-judicial Bodies · NCERT: Beyond NCERT
Meaning
An AI Safety Institute (AISI) is a public body that studies and tests artificial intelligence (AI) systems to find safety risks, meaning ways an AI could cause harm. India's version is proposed in the India AI Governance Guidelines (November 2025), issued by the Ministry of Electronics and Information Technology (MeitY) under the IndiaAI Mission. It comes from a policy document, not from an Act of Parliament, so it is not a statutory body (one created by a law). It matters because it gives India a dedicated technical body to check AI harms, instead of leaving safety only to the companies that build AI.
Example
The India AI Governance Guidelines propose three new bodies together: an AI Governance Group, a Technology & Policy Expert Committee and an AI Safety Institute. Of the three, the AI Safety Institute is the one meant to handle safety risks. Experts say it should record ordinary, everyday harms too, such as people trusting chatbots too much, not only dramatic future risks.
Don't confuse with
- AI Governance Group: also proposed in the same Guidelines, but its job is to coordinate and steer overall AI policy. The AI Safety Institute's job is technical safety work.
- Statutory regulator (like SEBI or TRAI): an Act of Parliament creates it and gives it legal powers to license, make rules and impose penalties. A proposed AI Safety Institute has no Act behind it and no such powers.