The Financial Stability Board (FSB) has published its final report on Sound Practices for the Responsible Adoption of Artificial Intelligence, setting out twelve sound practices designed to help financial institutions harness the benefits of AI while managing its risks in ways that support the resilience of the financial system.
The report, released on October 15 along with an overview of the public consultation responses it received, concludes a process that began when the FSB floated a consultative version of the practices in June 2026. As Reuters reported, the final text places particular emphasis on effective human oversight as banks and other institutions begin deploying increasingly autonomous agentic AI systems.
What the 12 Sound Practices Cover
The FSB organized its recommendations into two groups. The first emphasizes organization-wide AI governance and highlights four priorities: the pivotal role of boards and senior management in setting and overseeing the alignment between a financial institution's AI adoption and its business model, risk appetite and strategy; maintaining a governance framework with well-defined roles and responsibilities; incorporating AI risks into the overall risk management framework with effective documentation of AI use; and undertaking continuous learning and adaptation as the technology evolves.
The second group sets out a comprehensive framework for managing risks across the AI lifecycle, from development through deployment. It calls for systematic assessment of the materiality and risks of each AI use case at inception and thereafter, careful selection of AI technologies, appropriate data governance, explainability of different types of AI alongside compensating controls where explainability falls short, ongoing evaluation of use-case performance, appropriate and effective human oversight, management of AI-related cyber and ICT risks, and management of risks from third-party AI dependencies.
A Toolkit, Not a Rulebook
The FSB is careful to frame the practices as a practical toolkit rather than binding regulation. Institutions are expected to tailor the practices to their own context and risk profile and integrate them into existing frameworks, including those already in place for cyber and third-party risk. The report includes case studies illustrating the benefits financial institutions have reported from AI adoption and how some of the sound practices have been applied proportionately.
The practices also aim to promote coordination, cooperation and information-sharing among financial institutions, supervisors and other stakeholders, both within and across jurisdictions. The FSB notes that they build on, and are broadly compatible with, existing and ongoing work by the FSB and other standard-setting bodies — an important point for global banks that would otherwise face a patchwork of conflicting AI rules.
As AI technologies and applications evolve, the FSB says it will monitor developments and update the sound practices as necessary.
Why Human Oversight Is the Flashpoint
The emphasis on human oversight reflects where the industry's anxiety has shifted. Financial firms are moving from AI as an analytics tool toward agentic systems that can execute multi-step tasks — researching counterparties, drafting compliance filings, interacting with customers — with limited supervision. Reuters reported that the FSB's message is that such systems need more human oversight, not less, as their autonomy grows.
That concern is not abstract for supervisors. The FSB's chair, Bank of England Governor Andrew Bailey, has repeatedly warned this year about the financial-stability implications of frontier AI, including concentration risk if a handful of model providers and cloud platforms become critical infrastructure for the sector. The final report's governance-first structure — putting board accountability before technical controls — is a direct answer to that worry.
What Happens Next
The FSB coordinates financial-stability work across national authorities and international standard-setting bodies, bringing together members from 24 countries and jurisdictions while conducting outreach to roughly 70 more through its six Regional Consultative Groups. Its secretariat is based in Basel at the Bank for International Settlements.
National regulators are expected to reference the sound practices as they finalize their own AI rules for banks, insurers and asset managers. For financial institutions, the practical takeaway is that AI governance is moving from ad hoc experimentation to documented, board-level accountability — and that supervisors will be asking for evidence.
Stay Ahead of AI
Track every AI regulation, from Basel to Brussels — read more AI news on AI Buzz Wire.
Stay Ahead of AI →
