Artificial intelligence now sits at the top of the Financial Stability Board’s risk message to the G20. In his letter to G20 finance ministers and central bank governors ahead of the Asheville meeting, FSB Chair Andrew Bailey wrote that “frontier AI may have the ability materially to alter the speed, scale and economics of cyber risk.” At the same time, the FSB is finalizing a set of sound practices for how financial institutions adopt AI. Both files will shape what supervisors expect of banks, insurers, asset managers and their technology vendors.
The FSB chair’s warning
Bailey’s letter describes the emergence of frontier AI models, with growing autonomy and problem-solving ability, as a complicating factor in an already fragile risk picture. The FSB says the letter warns that markets remain vulnerable to a potentially disorderly correction that could spread across borders, pointing to fragilities in sovereign debt markets, vulnerabilities in private credit and stretched asset valuations. The FSB’s May 2026 report on private credit put that sector at an estimated $1.5 to 2 trillion in assets.
On AI, the letter says supporting safe and responsible release and deployment of frontier models on a global basis is a priority, and that the FSB is examining issues around their safe deployment for cyber defense in the financial sector. It notes that many jurisdictions lack protocols for managing how such models are developed, released and deployed.
The sound practices
The FSB published a consultation report on June 10, 2026, setting out “a menu of 12 sound practices” for responsible AI adoption by all types of financial institutions. The practices cover organization-wide AI governance and the stages of AI development and deployment, with boards and senior management as a key audience. The consultation asked whether the practices strike the right balance for “emerging and new complex forms of AI, such as GenAI and agentic AI.” Comments closed on July 22, and the FSB published the responses on August 6.
G20 finance ministers signaled support in the Asheville chair’s statement: “We look forward to finalization of the FSB’s paper on Sound Practices for Responsible Adoption of AI.” The same text describes AI as “a general-purpose technology” and welcomes “leveraging AI-enabled innovation to strengthen cyber resilience.” The FSB has not published a date for the final paper.
What to watch
The next checkpoint is the G20 finance ministers’ meeting in Bangkok on October 15, on the margins of the IMF and World Bank Group Annual Meetings, followed by the leaders’ summit in Miami on December 14–15. Companies should watch whether the final sound practices tighten expectations on third-party AI providers or agentic systems after consultation, and whether the FSB’s frontier AI work produces specific guidance on model release and cyber defense. The sound practices are not binding. Their effect will depend on how national supervisors fold them into existing rules on governance, model risk and operational resilience.
Sources
- Financial Stability Board: FSB Chair’s letter to G20 Finance Ministers and Central Bank Governors, August 2026
- Financial Stability Board: Letter text (PDF)
- Financial Stability Board: Sound Practices for Responsible Adoption of AI, consultation report, June 10, 2026
- U.S. Department of the Treasury: G20 Chair’s Statement, Asheville, September 1, 2026
- G20 (U.S. presidency): Events calendar
- Financial Stability Board: Publications















