The Financial Stability Board on Wednesday published a consultation report setting out sound practices for the responsible adoption of artificial intelligence by financial institutions. The report proposes 12 sound practices spanning organisation-wide AI governance and lifecycle risk management.
In its release, the FSB said the recommendations were designed to support boards and senior management in shaping business strategy and technology-adoption decisions, and to address risks that may be introduced or amplified by the rapid pace of AI adoption across the sector. The report is aimed at bringing greater consistency to how supervised institutions organise themselves around AI.
TWELVE SOUND PRACTICES
The 12 practices set out expectations across two broad areas. The first covers organisation-wide AI governance, including the roles and responsibilities of the board and senior management, alignment of AI activity with the institution's overall risk appetite, and integration into existing enterprise risk-management frameworks. The FSB stresses that AI-specific governance should sit alongside, rather than replace, conventional model risk and technology governance.
The second area addresses the AI lifecycle, from data acquisition and model development to deployment, monitoring and retirement. The lifecycle framing recognises that risks evolve as models move from experimentation into production, requiring controls that are adapted to each stage. Ongoing monitoring for model drift, performance degradation and unintended behavioural change receives particular attention in the recommendations.
GENERATIVE AND AGENTIC AI IN SCOPE
The report explicitly covers generative AI and agentic AI, the latter referring to systems capable of taking autonomous actions to achieve defined goals with limited human oversight. The FSB flagged the risks that may be introduced or amplified by rapid adoption of these technologies, including new forms of operational, model, third-party and information-security risk. Concentration in a small number of foundational model providers is also identified as a potential source of systemic vulnerability.
Case studies from real-world financial-institution implementations are included in the consultation, providing practical illustrations of how the sound practices can be applied. The FSB has increasingly used such examples in its work to bridge the gap between high-level principles and day-to-day supervisory expectations, and the case studies are drawn from institutions across banking, insurance and asset management.
The consultation is aimed at national authorities, industry associations, financial institutions and other interested parties. Comments will feed into the FSB's final report on the topic. A virtual outreach event has been organised to discuss the sound practices with stakeholders. The consultation document and details of the outreach event are available on the FSB website, and industry bodies have already flagged their intention to submit coordinated responses. The FSB's work on AI complements parallel initiatives at the Basel Committee on Banking Supervision, the International Association of Insurance Supervisors and IOSCO, which have each taken up aspects of the same topic. Convergence across those bodies will be closely watched by internationally active financial institutions, which face the prospect of overlapping expectations from prudential, conduct and market regulators simultaneously. The FSB is chaired by Klaas Knot and coordinates the work of national authorities and international standard-setters, with its output typically feeding through into supervisory expectations across member jurisdictions in the months following a final report.