Key facts
- RBI wants banks to add kill switches to AI models
- Human oversight mechanisms also required under proposed framework
- Applies to AI deployed in Indian banking and financial services
The Reserve Bank of India wants banks to build kill switches and human oversight controls into their AI systems, signalling that India's central bank is moving toward formal guardrails on how artificial intelligence is deployed within the financial sector.
The RBI's position, as reported by MediaNama, reflects a global trend of financial regulators demanding that AI systems used in high-stakes environments — credit decisions, fraud detection, customer interactions — remain under meaningful human control. A kill switch would allow banks or regulators to immediately disable an AI model if it begins producing harmful or erroneous outputs, a safeguard that has become standard in responsible AI deployment frameworks abroad.
For Indian banks, which have been rapidly integrating AI into everything from loan underwriting to customer service chatbots, this could require significant re-engineering of existing systems. Many current AI deployments operate with limited real-time human supervision, and retrofitting oversight mechanisms is both technically challenging and expensive.
The RBI's intervention also comes at a moment when AI errors in financial services — biased credit denials, incorrect fraud flags, or automated errors at scale — carry serious consumer protection implications. A single faulty AI model operating across millions of accounts could cause widespread harm before a human review process catches it.
No formal circular has been issued at this stage, but the RBI's stated intent is a strong signal to banks to begin preparing their AI governance frameworks now. Compliance timelines and specific technical requirements are expected to be detailed in forthcoming guidance.