Governance Frameworks for the Age of Artificial Intelligence
Financial institutions are recalibrating their operational strategies as global regulators finalize new standards for the responsible deployment of artificial intelligence. This document provides a structured approach for firms to harness the benefits of machine learning and data processing while mitigating systemic risks.
Artificial intelligence is defined as the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, and problem-solving. While the field was founded as an academic discipline in 1956, it has traversed multiple cycles of optimism and periods of reduced funding known as AI winters. Growth accelerated significantly after 2012 when graphics processing units began to be used to train neural networks, and the subsequent emergence of transformer architecture in 2017 further expanded the technical possibilities. The current decade has seen a surge in interest driven by generative AI, which allows systems to create and modify media. Companies including OpenAI, Google DeepMind, and Meta are currently pursuing the development of artificial general intelligence, which aims to complete nearly any cognitive task at the level of a human.
Establishing Institutional Oversight
The Financial Stability Board identifies 12 sound practices to help financial institutions integrate these technologies safely. The first group of these practices focuses on organization-wide governance. According to the Financial Stability Board, senior management and boards of directors play a primary role in aligning AI adoption with the specific business models, risk appetites, and strategies of their institutions. This governance structure requires well-defined roles and responsibilities, the integration of AI-related risks into broader risk management frameworks, and the maintenance of effective documentation for all AI use cases.
The board also emphasizes the necessity for continuous learning and adaptation within financial organizations. By establishing these internal controls, firms can ensure that their adoption of new technology remains consistent with their overall operational resilience. The Financial Stability Board encourages leadership teams to use these practices as a toolkit that can be tailored to the specific context and risk profile of their institution. This approach allows for the integration of AI management into existing frameworks, such as those already used for cyber and third-party risk management.
Managing Risk Across the AI Lifecycle
The second group of sound practices addresses the technical and operational risks that emerge throughout the AI lifecycle, from initial development to full deployment. The Financial Stability Board highlights the importance of assessing the materiality and risks of AI use cases at the inception phase. This includes the selection of appropriate technologies, the implementation of data governance, and the establishment of explainability standards for various AI models. When models are complex, the report suggests the use of compensating controls to manage uncertainty.
Human oversight remains a central component of the framework. The Financial Stability Board requires that institutions maintain effective human supervision to evaluate the performance of AI systems regularly. Furthermore, the framework addresses the management of cyber and information and communication technology risks, as well as the risks associated with third-party AI dependencies. By focusing on these areas, financial institutions can better manage the challenges that arise when integrating automated systems into sensitive financial processes.
Future Coordination and Market Adoption
The adoption of these sound practices is intended to promote coordination, cooperation, and information-sharing among financial institutions, supervisors, and other stakeholders across different jurisdictions. As AI adoption increases, the ability of regulators to share insights becomes more important for maintaining the stability of the global financial system. The Financial Stability Board report builds on existing work and is designed to be compatible with ongoing regulatory efforts worldwide.
The shift toward these structured governance models comes as the broader implications of AI are being examined across society. Recent observations suggest that the speed and availability of AI tools are changing how users interact with information. While these tools can provide patient, well-calibrated explanations for complex topics, they also change the nature of how individuals approach uncertainty. The capacity to sit with not-knowing is a developmental achievement that supports creative thinking and emotional regulation. As financial institutions move forward, the challenge will be to balance the efficiency of AI-driven decision-making with the need for deliberate, human-led oversight. The sound practices published by the Financial Stability Board provide the necessary structure for firms to navigate this transition while maintaining the integrity of their operations.
Practical Application and Case Studies
To assist firms in implementing the 12 sound practices, the Financial Stability Board included case studies within its October 2026 report. These examples illustrate the tangible benefits reported by various financial institutions and demonstrate how organizations have applied the board’s recommendations in a proportionate manner. By reviewing these real-world applications, leadership teams can better understand how to tailor the toolkit to their specific institutional context and risk profile.
The integration of these practices is intended to support the resilience of the global financial system while allowing firms to harness the advantages of machine learning and data processing. Because the sound practices are designed to be compatible with existing frameworks, such as those already utilized for cyber and third-party risk management, they provide a flexible structure for firms. This approach ensures that as institutions adopt new technologies, they maintain the ability to manage risks throughout the entire lifecycle of their AI systems, from the initial development phase through to full deployment.
Sources
- Artificial intelligence
- FSB finalises sound practices for the responsible adoption of artificial intelligence (AI)
- What My Son Is Learning From Artificial Intelligence
Photo: Federalreserve (Public domain)
