The Reserve Bank of India has released draft guidance mandating banks and regulated entities to implement kill switches and ensure robust human oversight over AI and machine learning models to mitigate associated risks.
- Mandatory kill-switch mechanisms for AI system override
- Board-approved model risk management framework required
- Human oversight must guard against automation biases
What happened
The Reserve Bank of India (RBI) has published draft guidelines titled 'Guidance on Regulatory Principles for Model Risk Management, 2026' for public comment until July 24, 2026. This framework targets banks, NBFCs, co-operative banks, and other financial institutions deploying AI and statistical models in their operations. The draft mandates that all AI systems include kill switch mechanisms enabling independent override, suspension, or deactivation to manage emergency risks effectively.
In addition to kill switches, the draft requires financial entities to develop a comprehensive Model Risk Management Framework approved by their Boards. This framework should cover model governance, classification, documentation, validation, and ongoing monitoring. Significant emphasis is placed on ensuring human supervision over AI systems to mitigate issues like automation bias and over-reliance on AI outputs. The guidance also highlights the need for periodic review of AI models' risk tiers and third-party dependencies.
Why it matters
AI and machine learning adoption in India’s financial sector has accelerated, driven by digital transformation, advanced analytics, and greater use of third-party AI providers. While these technologies improve customer service and risk management, they also introduce complex risks related to autonomy, reliability, and data integrity. Without robust controls, automated models could result in erroneous or biased decisions that harm customers and the broader financial ecosystem.
The RBI’s guidelines address these concerns by imposing a regulatory framework designed to enhance accountability, oversight, and transparency for AI models. Having kill switches ensures financial institutions can quickly intervene in the event of model failure or cyber threats. The stress on human oversight aims to preserve control over automated decisions and prevent blind reliance on AI outputs, which can compromise operational and reputational stability.
What to watch next
Stakeholders, including banks, NBFCs, and technology providers, should prepare to submit feedback on the RBI’s draft guidelines by the July 24 deadline. The final regulations will likely influence AI adoption strategies, risk management policies, and technology investments in India’s financial sector. Operational readiness to implement kill switches and strengthen governance will be critical for compliance.
Investors, fintech firms, and regulators should monitor how these rules evolve and their impact on innovation and competitiveness. RBI may issue further AI-specific requirements over time, reflecting technology development and emerging risks. Additionally, how institutions balance AI-driven efficiency gains with regulatory demands for transparency and human oversight will shape the future landscape of AI in Indian finance.