Goldman Sachs India faces tighter RBI rules on AI risk oversight

The Reserve Bank of India put banks on notice with a June 24 draft that would require board-approved oversight for every model in use, including AI and machine-learning systems, and it is taking comments until July 24. The proposal reaches third-party models too, and it adds independent validation, human oversight for automated decision-making and extra cybersecurity controls for customer-facing generative AI. For banks, that shifts AI from an experiment to a documented control environment with inventories, escalation paths and clearer accountability.

For Goldman Sachs employees in India, the timing lands in a footprint the firm says is now central to its global operations. Goldman says its Bengaluru office opened in 2004 and Hyderabad in 2021, both as technology hubs integral to global businesses and operations, while Mumbai remains its onshore India base. The firm says those two cities now house about 9,000 professionals, making India its second-largest presence globally. That makes the RBI's draft more than a regulatory footnote: it reaches the engineers, risk managers, compliance staff, operations teams and lawyers who would have to prove who owns a model, who validates it and who signs off before it goes live.

The central bank has been building toward this for a while. Its FREE-AI report, dated August 13, 2025, followed a committee formed in December 2024 to examine responsible AI in financial services, and the report laid out 26 recommendations under six strategic pillars. It also included a suggested board policy on AI and an AI incident reporting form, which signals that the RBI wants governance machinery, not loose principles, around the technology. For Goldman and its peers, that is likely to mean more vendor due diligence, tighter model inventory management and more work for control functions before AI tools can move from pilot projects into production.