Banks must maintain AI inventories and human oversight, RBI says
The Reserve Bank of India has outlined supervisory expectations for artificial intelligence adoption across domestic banks at the FIBAC 2026 conference. The central bank mandated board-level governance, system explainability, and full human oversight.
No refuge behind algorithms
The Reserve Bank of India (RBI) established immediate supervisory priorities for artificial intelligence integration, requiring lenders to maintain full inventories of all deployed systems.
The central bank stressed that algorithms cannot replace managerial responsibility, noting that “the model decided” will not be accepted as justification for credit denials or automated outcomes.
Under the Framework for Responsible and Ethical Enablement of AI, institutions must conduct periodic stress testing and red-teaming exercises.
The regulator also instructed boards to implement explainability mechanisms for retail credit assessments, address third-party vendor dependencies, and eliminate algorithmic bias.
Public digital rails and Basel deadlines
The supervisory push coincides with broader regulatory reforms, including the scheduled implementation of all applicable Basel III guidelines starting on April 1, 2027.
The RBI highlighted public digital infrastructure such as the Unified Lending Interface and Account Aggregator rails to expand credit access for micro, small, and medium enterprises.
To safeguard payment networks, the central bank continues developing sector-wide utilities, including the MuleHunter tool and the Digital Payments Intelligence Platform.
Accountability cannot be outsourced
The RBI rightly rejects tech-evangelism by placing legal liability squarely on bank directors.
Enforcing explainability will curb reckless model deployment across vulnerable retail borrowers.
However, supervising complex third-party software across smaller regional lenders will test regulatory capacity.