The Financial Stability Board (FSB), the international body behind global financial stability standards, has released the final version of its guidance on responsible artificial intelligence adoption.
The guidance sets out 12 sound practices that place board-level accountability and lifecycle risk controls at the centre of how financial institutions deploy AI.
The final report follows a consultative draft issued in June 2026, and its contents have been shaped by the public responses gathered on that earlier version. The 12 practices are designed to let financial institutions capture the advantages of AI while keeping its risks in check, so that wider adoption strengthens rather than undermines the resilience of the financial system.
The FSB has split the practices into two categories. The first focuses on governance across the whole organisation. It assigns the board and senior management a central role in making sure AI adoption fits the institution’s business model, risk appetite and strategy, and in monitoring that fit over time.
This category also calls for a governance framework with clearly allocated roles and responsibilities, the inclusion of AI risks within risk management frameworks alongside thorough documentation of how AI is used, and an ongoing commitment to learning and adaptation.
The second category lays out a full framework for handling risk across the AI lifecycle, from development through to deployment. Here the FSB calls for systematic assessment of how material and risky each AI use case is, both at the outset and on a continuing basis. Further practices cover choosing suitable AI technologies, sound data governance, the explainability of different forms of AI and the use of compensating controls where needed, performance evaluation of AI use cases, and meaningful human oversight.
The final two practices address AI-related cyber and information and communication technology (ICT) risks, as well as exposure created by reliance on third-party AI providers.
The FSB presents the practices as a hands-on toolkit rather than a rigid rulebook. Institutions can adapt them to their own circumstances and risk profile and fold them into frameworks already in place, including those covering cyber and third-party risk.
The report also features case studies showing benefits that financial institutions have reported and examples of proportionate application of certain practices.
Beyond individual firms, the practices are intended to encourage coordination, cooperation and information-sharing between financial institutions, supervisors and other stakeholders, both domestically and internationally.
They draw on, and are broadly aligned with, completed and ongoing work by the FSB and other standard-setting bodies. The FSB has said it will track how AI technologies and their applications develop and revise the practices where necessary.
Copyright © 2026 FinTech Global
