RBI Proposes AI Kill Switch For Banks Under Model Risk Framework
The central bank has proposed draft rules requiring banks and other regulated entities to immediately suspend or deactivate AI models if they malfunction or pose risks.
News
Image Credit- Chetan Jha/ MIT Sloan Management Review India
The Reserve Bank of India (RBI) has proposed a framework to govern the use of artificial intelligence and other decision-making models in the financial sector, including a requirement that regulated entities be able to immediately disable AI systems if they malfunction or pose risks.
Released for public consultation on Wednesday, the draft Model Risk Management Framework proposes that regulated entities maintain mechanisms to override, suspend or deactivate AI models whenever their outputs become harmful, inaccurate or unreliable.
The proposal covers the full spectrum of models used by banks and financial institutions, from basic analytical models to advanced artificial intelligence and machine learning systems.
The RBI said the rapid expansion of model-based decision-making across financial services has increased the need for stronger governance and oversight.
“Considering model usage has expanded significantly and regulated entities are increasingly using models, including those employing artificial intelligence/machine learning, across various business and decision-making processes, weaknesses in their governance, oversight, risk management and controls may expose the regulated entities to financial, operational, compliance, and reputational risks,” the central bank said in the draft.
The regulator also warned that poor model governance could result in “inaccurate outcomes, flawed decisions, financial losses, operational disruptions, compliance failures and other adverse consequences for entities, consumers and the financial system.”
Under the proposed framework, banks would remain accountable for every model they deploy, regardless of whether it was developed internally or procured from an external vendor. The draft also requires institutions to conduct due diligence before adopting third-party models.
The RBI has proposed that every AI-assisted decision remain subject to human oversight, aiming to reduce the risk of employees relying unquestioningly on automated recommendations.
The framework introduces a risk-based classification system under which banks must assess models according to their potential impact and apply governance measures proportionate to that risk. High-risk models would require approval from the Board’s Risk Management Committee before deployment, and their risk ratings would need to be reviewed at least annually.
If a model’s risk exceeds an institution’s approved threshold, the RBI said banks should take corrective measures that could include strengthening controls, limiting its use, fixing identified issues or retiring the model altogether. Such cases would also need to be reported to the board’s risk committee.
The draft places model governance directly under board oversight, requiring every regulated entity to establish a board-approved Model Risk Management Framework covering internally developed models, externally sourced systems and hybrid deployments.
The RBI also highlighted emerging risks linked to AI adoption, including dependence on a limited number of AI providers and the possibility that AI systems could introduce new cybersecurity vulnerabilities if not properly managed.
For AI-powered customer services, the regulator proposed that banks clearly inform customers whenever they are interacting with an AI system and provide an option to speak with a human representative.
The central bank has invited comments on the draft framework until July 24 before issuing the final guidelines.
