On 10 June 2026, the Financial Stability Board (FSB) published a consultation report titled 'Sound Practices for Responsible Adoption of Artificial Intelligence (AI).' The report sets out 12 nonbinding sound practices derived from supervisory observations and regulated institution experience across FSB member jurisdictions. The FSB closed the public comment period on 22 July 2026, published consultation responses on 12 August 2026, and will publish the final report in October 2026.
The 12 sound practices fall into two groups. Sound practices 1 to 4 address organisation-wide AI governance: board-level accountability for AI risk, integration of AI risk into enterprise risk management, clear internal ownership of AI decisions, and documented risk appetite for AI adoption. Sound practices 5 to 12 address AI lifecycle management: training data governance, model risk management, explainability calibrated to the stakes of the AI decision, meaningful human oversight, third-party AI vendor oversight, and specific governance obligations for generative and agentic AI systems operating with reduced human intervention. The FSB states that the practices do not constitute a binding international standard and were not designed to govern frontier AI model risks, though several practices provide relevant governance tools for those risks.
Banks, asset managers, insurers, clearing houses, payment service providers, and market infrastructure operators in FSB member jurisdictions are the FSB's intended audience. Member jurisdictions include the G20 economies, the European Union, and major international financial centres. National supervisors are expected to draw on the sound practices when developing or updating domestic AI governance expectations for regulated institutions. Institutions subject to the Basel Committee's Principle 15 on AI risk management, the IOSCO AI guidance, or the IAIS draft AI principles will find the FSB practices extending and coordinating those prior documents.
The FSB acknowledges that no member jurisdiction has adopted a statute governing AI in financial services end-to-end, leaving institutions subject to sector-specific supervisory expectations across banking, securities, and insurance. The October 2026 final report will incorporate consultation feedback and may refine the scope or content of individual practices. Consultation responses are publicly available on the FSB website.
Licentium advises banks, payment institutions, asset managers, and fintech companies on AI governance and regulatory compliance in financial services across EU, UK, and international jurisdictions. Work we undertake includes AI governance policy design, model risk management assessment, third-party AI vendor due diligence, regulatory horizon scanning, and multi-jurisdictional AI compliance programme development.