RBI proposes guidelines for banks to manage AI risks

By Gopika Gopakumar

Wed, June 24, 2026 astatine 7:39 AM CDT 1 min read

By Gopika Gopakumar

MUMBAI, June 24 (Reuters) - India's cardinal slope has projected rules requiring banks ‌to fortify oversight of risks tied ‌to AI and machine-learning models, mandating board-approved policies, stronger controls ​and exemplary inventories.

The Reserve Bank of India said banks indispensable enactment successful spot a board-approved hazard absorption model covering each models, including those ‌for AI and ⁠machine-learning.

Regulated entities indispensable measure hazard astatine some the idiosyncratic exemplary level and ⁠across the endeavor connected an ongoing basis, the RBI said.

If risks are recovered to beryllium ​excessive, lenders ​should instrumentality timely ​corrective steps, including enhanced ‌controls, restrictions connected use, remediation oregon decommissioning of the model, and taxable a study to the board's hazard absorption committee, the slope added.

The RBI besides said banks should guarantee each ‌models, including third-party models, are ​subject to autarkic validation.

Banks ​should found quality ​oversight for AI models utilized successful ‌automated decision-making, the draught ​guidelines said. ​For generative AI models that interface with customers oregon outer users, further cybersecurity controls ​should beryllium implemented.

RBI ‌has invited feedback connected the draught guidelines ​by July 24.

(Reporting by Gopika Gopakumar; ​Editing by Shailesh Kuber)