RBI Draft Rules: Banks Can't Pass the Buck to AI for Lapses

RBI Draws a Line on AI Accountability

The Reserve Bank of India has released a draft framework that puts the onus squarely on banks for any decisions made by artificial intelligence systems. The message is unambiguous: lenders cannot hide behind the algorithm.

The guidelines, published by the central bank on Friday, stress that regulated entities remain fully responsible for outcomes generated by AI or machine learning tools. This includes credit underwriting, fraud detection, customer service, and risk management.

No Room for the 'Algorithm Did It' Defence

RBI’s draft explicitly states that banks cannot cite technical errors or biases in AI models as a justification for violating regulatory norms. If a loan is denied unfairly or a transaction flagged incorrectly, the bank must answer for it — not the software vendor or the code.

This marks a significant shift in tone. While many lenders have adopted AI to speed up processes, the regulator is now ensuring that accountability remains human. The draft requires banks to maintain a clear audit trail of all decisions made by algorithms.

Transparency and Explainability Become Mandatory

Under the proposed rules, banks will need to document how their AI models arrive at specific decisions. This includes explaining why a loan application was rejected or why a certain interest rate was offered. The goal is to prevent black-box algorithms from operating without oversight.

Lenders must also ensure that AI systems do not discriminate against borrowers based on caste, religion, gender, or region. The draft calls for regular testing of models to detect and correct bias. Regulators will have the power to demand explanations for any decision flagged by customers.

What This Means for Banks

For India’s banking sector, which has rapidly adopted AI for everything from chatbots to credit scoring, the draft introduces new compliance burdens. Banks will have to invest in explainable AI tools and hire staff who can interpret model outputs. Smaller lenders may find this challenging.

Industry observers note that the RBI is following global trends. Regulators in the European Union and the United States have also moved to hold financial firms accountable for AI-driven outcomes. India’s draft aligns with this push for responsible AI governance.

The RBI has invited comments on the draft from stakeholders. The final guidelines are expected later this year. Banks will likely get a transition period to comply, but the direction is clear: algorithms are tools, not scapegoats.

What to watch next: how banks redesign their AI oversight frameworks and whether the RBI introduces similar rules for non-bank financial companies and fintech firms.