The Financial Stability Board issued final guidance on October 15 for financial institutions using artificial intelligence. The guidance calls for stronger human oversight of AI agents that can plan and perform tasks with limited supervision.
FSB sets 12 practices
The global financial standard setter published its final Sound Practices for Responsible Adoption of Artificial Intelligence. The framework contains 12 practices covering board oversight, defined responsibilities, risk assessment, data governance, explainability, performance evaluation, cybersecurity and third-party AI dependencies.
The FSB said financial institutions can tailor the practices to their size, activities and risk profile. It strongly encouraged boards and senior managers to use them when developing AI governance, risk management and control frameworks.
Agents need human approval
The guidance gives special attention to agentic AI, which can plan and execute tasks rather than only generate text or predictions. Reuters reported that the FSB strengthened its recommendations for these systems after considering feedback and growing interest from financial institutions.
For higher-risk uses, the FSB says more intensive human oversight will be critical. It also says firms should establish risk-based triggers that require an agent to seek human approval.
The final guidance removes language that compared AI systems with “synthetic employees.” Instead, it says firms should recognize that AI agents do not reason like people and may interpret predefined goals in unexpected ways.
Global firms face closer scrutiny
The FSB said the practices are intended to improve coordination, cooperation and information-sharing among financial institutions, supervisors and other stakeholders as AI adoption grows.
The board will monitor changes in AI and update the practices when needed. The framework covers how institutions assess AI use cases, manage outside providers and maintain human oversight throughout the technology’s lifecycle.
References
Financial Stability Board. (2026, October 15). FSB finalises sound practices for the responsible adoption of artificial intelligence (AI). https://www.fsb.org/2026/10/fsb-finalises-sound-practices-for-the-responsible-adoption-of-artificial-intelligence-ai/
Reuters. (2026, October 15). Global financial watchdog calls for more human oversight of agentic AI. https://live.euronext.com/en/financial-news/global-financial-watchdog-calls-more-human-oversight-agentic-ai