The AI Revolution in Banking
Indian banks have enthusiastically embraced Artificial Intelligence, and for good reason. AI promises to transform lending, improve customer service, detect fraud, and bring more people into the formal financial system. For customers, this means faster
loan approvals, 24/7 support from chatbots, and more personalised services. For banks, it means greater efficiency and the ability to assess risk more accurately. RBI Governor Sanjay Malhotra recently acknowledged that AI could define this decade for Indian banking just as decisively as digitalisation defined the 2010s. The technology offers a path to fundamentally change credit delivery, especially for those with limited financial history, by using alternative data like utility payments and digital footprints to assess creditworthiness.
Unpacking the System-Wide Risk
The RBI's core concern is not with AI itself, but with how it is being deployed. The potential for "system-wide" or "systemic" risk arises from what is known as concentration and herding. If many banks rely on the same few AI models, data sources, or technology vendors, a single flaw, bias, or cyberattack could trigger a cascading failure across the entire financial system. Imagine if a popular credit-scoring model was found to have a critical error, leading multiple banks to simultaneously misprice risk or cut off lending to a specific sector. This could create financial instability that goes far beyond a single institution. Governor Malhotra flagged this as a key risk, noting that similar AI-based trading models could cause banks to behave identically during market stress, amplifying volatility instead of containing it.
The 'Black Box' Dilemma
One of the most significant challenges with advanced AI is the 'black box' problem. Many complex models arrive at conclusions through processes that are not easily understood by their human creators. When an AI system denies a loan, for example, the bank needs to be able to explain why. The RBI has been firm on this point: "'The model decided' can never be an acceptable answer to a customer, an auditor, or the Reserve Bank," Governor Malhotra stated. This lack of transparency makes it difficult to check for hidden biases. An AI trained on historical lending data could unintentionally perpetuate and even amplify existing biases against certain communities or occupations, leading to new forms of financial exclusion.
Accountability in the Age of Algorithms
Ultimately, the RBI's main message is about accountability. The central bank has stressed that responsibility for any decision must lie with the bank itself, not with the algorithm. The governor warned that the biggest risk is the potential erosion of human judgment and accountability. To counter this, the RBI has urged banks to establish board-approved AI governance policies, maintain a complete inventory of all AI systems in use, and stress-test them just like any other material risk. Meaningful human oversight, with the ability to intervene and override AI-driven decisions, is being positioned as a non-negotiable principle.
The Path Forward: Regulation and Responsibility
The RBI's stance is not to stifle innovation but to ensure it happens responsibly. The central bank sees AI as a capability to be harnessed, not just a risk to be contained. This involves creating a robust framework for governance. Banks are now expected to build the capacity to explain every decision that materially affects a customer. This includes strengthening cybersecurity to defend against new AI-enabled threats and ensuring data privacy compliance goes beyond just meeting legal minimums. The regulator is encouraging the use of its regulatory sandbox to test new AI use cases in a controlled environment, signaling a collaborative approach to navigating this new technological frontier. This dialogue between the regulator and the industry will be crucial in balancing the immense potential of AI with the need to maintain financial stability.
