The Bank of Japan published its fiscal 2026 survey of 150 financial institutions, finding that more than 90% are using or trialing generative AI, including about 80% already using it. Adoption increased across all business types, particularly Regional Banks II, while applications expanded from administrative tasks into core operations involving customer information. Direct delivery of AI-generated outputs to customers remains limited, and most institutions assessed the technology as usable for business but in need of improvement. Institutions have advanced information controls, internal rules and human involvement in critical decisions, but many reported that governance, third-party risk management, and safety and security measures remain under development or need improvement. About half of the 136 institutions using or trialing generative AI permit confidential information, including customer data, to be processed for customer service-related operations. Priorities include business implementation, skilled personnel, IT infrastructure, data readiness and cybersecurity, particularly as firms consider customer-facing outputs, autonomous AI agents and frontier AI. The Bank of Japan expects senior management to oversee both the benefits and risks, secure necessary resources and strengthen risk management frameworks. It will continue addressing generative AI use and related risks through examinations, monitoring and seminars.