AI adoption requires board-driven risk governance, Malhotra warns

Reserve Bank of India Governor Sanjay Malhotra highlighted five core drivers for artificial intelligence in banking: expanding credit delivery via alternative data, enhancing customer service, driving financial inclusion through local voice interfaces, automating operations, and countering real-time API fraud.

Malhotra confirmed that India remains on target to implement all applicable Basel III guidelines on a calibrated glide path starting April 1, 2027.

Building on public infrastructure like Aadhaar, UPI, Account Aggregators, and the Unified Lending Interface, AI can make financial assessment instant.

However, the central bank expects institutions to maintain board-approved AI policies, audit trails, and mandatory explanation capabilities for automated credit or fraud outcomes.

The governor warned against seven major risks: black-box opacity, algorithmic bias, systemic concentration in foundation models, third-party vendor dependence, privacy breaches, adversarial cyber threats, and the erosion of human accountability.

Drawing on recommendations from the RBI's FREE-AI Committee and draft guidelines on Model Risk Management, Malhotra stressed that algorithmic decisions cannot bypass institutional responsibility.

Banks must preserve meaningful human oversight, conduct rigorous red-teaming, and maintain complete inventories of active models.