In a speech at a Cambridge University conference, Fernando Restoy, Chair of the Financial Stability Institute at the Bank for International Settlements, argued that supervisors must look beyond banks’ use of artificial intelligence and assess how AI-driven economic change affects their operational and strategic resilience. Prudential oversight should examine both the reduced time available to respond to AI-enabled cyber incidents and the longer-term effects on borrowers, credit portfolios and bank business models. Restoy called for model risk management guidance to be updated for advanced AI, including possible trade-offs between explainability and performance where risks are properly assessed and managed. He also urged supervisors to test banks’ incident response, service continuity and third-party recovery capabilities, while using forward-looking business model reviews, scenario analysis and horizon scanning to identify strategic vulnerabilities. Qualitative supervisory measures may be more effective than capital add-ons for firm-specific weaknesses, but supervisory judgment should operate within a transparent and consistent framework.