Regulation
India's central bank closed public consultation on its first AI-specific model risk framework on 24 July, replacing rules written in 2002
The Reserve Bank of India's draft Guidance on Regulatory Principles for Model Risk Management, released 24 June 2026, sets governance expectations for AI and machine-learning models across nearly the entire RBI-regulated financial sector, banks, NBFCs and beyond, and would retire a credit-risk-model framework that predates modern algorithmic trading entirely.
The detail worth sitting with is the baseline this replaces. The RBI's existing guidance on credit risk models dates to 2002, a framework written before India's algorithmic trading volumes existed in any meaningful form, let alone before AI-driven trading and investment tools. That a central bank overseeing one of the world's largest and fastest-growing retail trading markets has operated for over two decades without an AI-specific model risk framework says less about India's regulatory attentiveness and more about how recently AI adoption in financial models has outpaced every major regulator's rulebook, a pattern this site has now documented across the FCA, the FSB, the SEC and the CFTC in the space of a few weeks.
Scope is the other detail that separates this from a narrow trading-desk rule. The guidance applies across 'nearly the entire RBI-regulated financial sector,' banks and non-banking financial companies alike, which means it is not aimed specifically at algorithmic or AI trading desks, it is aimed at every material use of an analytical model, credit scoring, fraud detection, portfolio construction, trading signals, under one governance standard. That breadth is a deliberate choice: rather than write bespoke AI-trading rules the way the CFTC's Innovation Task Force or the UK's evolving sandbox approach have done, the RBI is folding AI model risk into the same validation, documentation and oversight discipline that already governs traditional credit models, extending an existing muscle rather than building a new one.
The 24 July consultation close lands this guidance at an unusually pointed moment. It arrives in the same week this site has covered a Zhejiang quant fund's crowded-factor drawdown and a formal alpha-decay model showing AI adoption compressing the market's price-discovery-versus-fragility tradeoff, both of which are precisely the kind of correlated, model-driven risk a model-risk-management framework is meant to catch before it becomes systemic. Whether the RBI's final guidance, once it moves from draft to binding rule, requires firms to actively monitor for correlated model behaviour across the sector, not just validate each model in isolation, will determine whether this framework is built for the AI-trading era or just relabelled for it.
Watch the finalised version for whether India's approach ends up closer to the FCA's outcomes-based, no-new-rules posture or to a more prescriptive validation-and-documentation regime in the mould of what ESMA has signalled for EU algorithmic trading. A market the size of India's retail and institutional trading base choosing one model over the other will carry weight for how other emerging-market regulators calibrate their own AI model risk rules over the next year.
Read the original: Vinod Kothari Consultants - India's central bank closed public consultation on its first AI-specific model risk framework on 24 July, replacing rules written in 2002. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.