DPDP Rules status: compliance window active. Read the update

Skip to content
AI Policy · Sector Draft

RBI Draft Model Risk Management Guidance, 2026

RBI's draft guidance on model risk for banks, including AI models. Released 24 June 2026. Comments closed 24 July 2026. Still a draft as of 6 October 2026.

Educational only · not legal adviceLast updated: 6 October 2026Reviewed by Sandesh Kokate, Editor

If you build or use AI for a bank, this document is for you. Its full title is the Draft Guidance on Regulatory Principles for Model Risk Management, 2026. The Reserve Bank of India's Department of Regulation issued it on 24 June 2026. The reference is Press Release 2026-2027/528, signed by Brij Raj, Chief General Manager.

Earlier RBI guidance set out principles. This draft goes further and spells out concrete operational expectations for banks that use AI models. It expects a way to shut a model off. It expects clear board-level ownership of model risk. And it expects banks to be able to explain the AI decisions they rely on.

Status: still a draft. The consultation window closed on 24 July 2026, so comments are no longer being invited. RBI has not issued final guidance and has not announced an implementation timeline. We last checked on 6 October 2026. Nothing here is binding today.

Once finalised, this guidance will replace Chapter 3 (Credit Risk Models) of the Guidance Note on Credit Risk Management dated 12 October 2002.

It sits alongside the FREE-AI Framework. Think of FREE-AI as the principles layer. This guidance is the operational layer, written specifically for banks.

In plain English: if you are a bank, or you build AI that banks use for decisioning, track this document. It is where the practical "how" of AI governance for Indian banking is being decided. You can no longer comment on it. Watch instead for the final guidance.

Source
RBI press release, 24 June 2026 — draft Guidance on Regulatory Principles for Model Risk Management

Draft — checked 2026-10-06. Opened RBI's press release of 24 June 2026 releasing the draft Model Risk Management guidance. Comments were invited by 24 July 2026.

Quick self-check

How ready are you?

Reading about AI governance expectations? Run through the readiness checklist to see what's already covered and what's missing.

Run it now