The Financial Stability Board (FSB) has published 12 sound practices designed to help financial institutions strengthen their governance and risk management as the use of artificial intelligence expands across the financial sector.
The final recommendations place particular emphasis on board-level accountability, human oversight, data governance and the management of risks arising from third-party AI providers.
The framework follows a consultation launched in June 2026 and is intended to help financial institutions incorporate AI-related risks into their existing governance and control arrangements, rather than establish entirely separate systems.
The FSB is encouraging boards and senior management teams to use the recommendations when reviewing how AI technologies are developed, deployed and monitored within their organisations.
BOARD ACCOUNTABILITY AND GOVERNANCE
The 12 practices are divided into two areas, beginning with organisation-wide governance and the responsibilities of senior management.
Boards are expected to oversee whether AI adoption is consistent with their institution’s business model, strategic objectives and appetite for risk. The framework also stresses the importance of clearly defined responsibilities, appropriate documentation and continuous assessment as AI capabilities develop.
Institutions are encouraged to incorporate AI risks into their wider risk management frameworks, ensuring that the technology is subject to appropriate scrutiny throughout its use.
The FSB also identifies continuous learning and adaptation as essential elements of governance, recognising that AI technologies and the risks associated with them are evolving.
MANAGING RISKS THROUGHOUT THE AI LIFECYCLE
The second part of the framework addresses the practical management of AI systems, from their initial selection and development through to deployment and ongoing monitoring.
Financial institutions are encouraged to assess the significance and potential risks of individual AI applications before introducing them and to review those assessments as circumstances change.
The recommendations cover the selection of suitable technologies, data governance, the evaluation of AI system performance and the ability to explain how systems produce their outputs.
Where explainability is limited, the FSB highlights the need to consider additional controls to manage the associated risks.
Human oversight is another central consideration, alongside the management of cyber security threats and information and communication technology risks.
The framework also addresses dependencies on external technology providers, reflecting the importance of third-party AI systems and services within financial institutions’ operations.
FLEXIBLE APPROACH TO IMPLEMENTATION
The recommendations are designed to be proportionate, allowing individual institutions to adapt them to their operations, existing controls and risk profiles.
The FSB has included case studies illustrating how financial institutions have applied some of the practices and the benefits they have reported.
Rather than prescribing a uniform approach, the report seeks to establish a common foundation for managing AI risks while supporting cooperation between financial institutions, regulators and other stakeholders across different jurisdictions.
The framework builds on existing work undertaken by the FSB and international standard-setting bodies, including initiatives covering cyber resilience and third-party risk management.
The FSB said it would continue monitoring developments in AI technology and its use within financial services, updating the sound practices where necessary.
The organisation, which is chaired by Bank of England governor Andrew Bailey, coordinates international work on financial stability and brings together national financial authorities, international institutions and regulatory bodies.
The final report, Sound Practices for Responsible Adoption of Artificial Intelligence, has been published alongside an overview of responses to the June consultation.
