AI framework charts next chapter for Indian banking, drives innovation

The FREE AI Committee Report heralds a new era for Indian banking by providing a framework to both harness and regulate artificial intelligence. The Reserve Bank of India (RBI) report presents a roadmap that highlights not only the opportunities enabled by AI, but also challenges and responsibilities for banks and financial institutions. 

The framework emphasises robust governance, clear accountability for AI, drafting of policy, risk management, and ethical guardrails. It calls for metrics to ensure fairness and diversity in AI outcomes, with banks expected to monitor, report, and actively mitigate algorithmic bias to ensure inclusivity. 

Under the framework, banks are encouraged to use AI to streamline operations, boost customer engagement, and strengthen risk management. AI models for banking can deliver faster loan processing, enhanced credit scoring, and agile regulatory compliance. The framework urges integration with India’s Digital Public Infrastructure (Aadhaar, Unified Payments Interface or UPI, and account aggregators) as a way to drive financial inclusion. For example, UPI transaction records and mobile usage data can be used as alternative sources of data, assessed using AI to increase the spread of credit availability. While the framework sets a progressive foundation, several critical areas require attention for clarity: 

Vendor liability: The allocation of responsibility between banks and AI vendors has not yet been fully established. Vendor responsibilities and obligations are addressed through regulated entities, and pre-market conformity assessments for vendors are currently not specified. Without robust model contracts and oversight, disputes could slow adoption and expose banks to compliance failures. A shared responsibility framework between banks, vendors, and regulators is essential if India is to avoid costly standoffs. Collaborative efforts may be required for contracts and oversight mechanisms. 

Data readiness: No AI strategy can succeed without reliable, high-quality data, and upgrading legacy systems for AI is challenging. Shifting datasets to Cloud is the first step, followed by meeting the Basel Committee on Banking Supervision 239 (BCBS 239) requirements such as strong data governance, lineage mapping, and data integrity, ultimately aiming for Single Source of Truth for AI. BCBS 239 is a set of 14 principles — 11 for banks and three for supervisors — aimed at strengthening banks’ risk management and decision-making. Banks and regulators need to collaborate to set more compelling standards and resolve issues of standardisation and cost-sharing, without which data readiness risks becoming a privilege of only the largest players. 

Sector-specific AI: The report’s vision for tailored AI models connected to public infrastructure presents an opportunity but also lists technical and privacy hurdles. Such models could deliver efficiency gains across banking, while supporting financial inclusion at scale. But this ambition comes with some hard questions: Who owns the models? How will data be shared and protected? And how will privacy be guaranteed when customer information flows through shared platforms? Dialogue and joint strategising are needed to realise this vision. 

Metrics for fairness: There is a need for banks and the RBI to define how fairness and diversity in AI outcomes should be measured, monitored, and reported across the sector. 

The RBI’s AI roadmap deserves credit for being both progressive and pragmatic. It marks a watershed moment for Indian banking by laying out a framework for the ethical and responsible use of AI and sends a clear signal: the financial sector must innovate, but with guardrails firmly in place. 

The framework’s success will be judged by how decisively banks act on it. Those who move first – by investing in data, clarifying accountability, and embedding ethical AI – will not just meet regulatory expectations, they will set the pace for innovation within the sector. 

Although numerous additional areas still require consensus, establishing clear priorities in these key domains will significantly contribute to shaping the future, enabling the banking sector and its regulators to support the development of a more transparent, trustworthy, and innovative AI ecosystem. 

 

The writer is partner, financial services risk consulting, EY India