Fed's Bowman Introduces the FSB's Draft 'Sound Practices' for Responsible AI Adoption: Governance Scaled to Materiality, a Lighter Touch for Low-Risk Uses and Comments Shaping a Final Report for the G20

On July 7, 2026, Federal Reserve Vice Chair for Supervision Michelle W. Bowman gave opening remarks at a virtual outreach event held by the Financial Stability Board (FSB), the international body that coordinates financial regulation among major economies. The event concerned the FSB's consultation report, Sound Practices for Responsible Adoption of Artificial Intelligence. Bowman chairs the FSB's Standing Committee on Supervisory and Regulatory Cooperation, which oversaw the work. This article is based on the text of her remarks as published by the Federal Reserve. It is important to be clear about what the remarks are: an introduction to an FSB consultation document, delivered in Bowman's capacity as committee chair, and not a new Federal Reserve rule or supervisory guidance. She said her views were her own and not necessarily those of the Federal Reserve Board or the Federal Open Market Committee.

The work behind the report

Bowman thanked Hern Shin Ho of the Monetary Authority of Singapore, who led the work stream, along with FSB Secretariat staff and colleagues at the U.S. Treasury and the Securities and Exchange Commission. The report is meant to help financial institutions adopt and use AI responsibly. It does so by describing how institutions, including banks, are already managing AI risks, with the stated goal of responsible innovation. It includes examples and in-depth case studies showing governance and controls that may suit different situations.

Practices, not prescriptions

A central message is that the report is not a checklist. Bowman said that the practices in it are not the only way to adopt and use AI responsibly. That matters because the document, as a collection of observed practices, could be read as an expected standard. Her remarks position it as a menu of approaches that institutions can adapt, and she asked respondents to say where the practices might be too prescriptive.

Materiality should set the level of control

The most concrete part of the remarks concerns how much governance an AI use should attract. Bowman said institutions should be specific about how they use AI and whether that use is material to their operations or to their legal and regulatory obligations. The actual use, and its materiality, should guide the type and intensity of governance and controls for each deployment. As an illustration of the principle, and not an example from the speech, a tool that helps draft internal summaries would warrant less oversight than one that makes credit decisions.

The corollary, in her words, is that lower-risk uses should receive a lighter supervisory and regulatory touch. That is a statement of approach from a senior supervisor, and it is worth noting that it was made about the FSB's consultation and not as a commitment by the Federal Reserve to specific supervisory changes.

Proportionality and smaller institutions

The report, Bowman said, focuses on proportionality. Practices that suit large, complex AI use may not fit smaller institutions with simpler uses. She said the report gives clear guidance to all institutions, including smaller ones, and she welcomed feedback on whether it strikes the right balance. She emphasised the importance of promoting innovation at financial institutions of all sizes, not just the largest. The remarks refer to smaller institutions in general and do not mention community banks by name.

The Fed's own experience

Bowman noted that the Federal Reserve has monitored banks' use of AI for nearly a decade and has seen increased use by banks of all sizes. The remarks contain no statistics, so this article cannot say how many banks use AI or for what.

What happens next

The FSB report is a consultation, so it can change. Bowman said feedback during the public comment period will inform the final report, which will be delivered to the U.S. G-20 presidency later in 2026. She asked respondents to say where the practices may be too prescriptive, where they miss differences in size, complexity or risk, where material risks may not be addressed, and where more clarity is needed.

What the remarks do not cover

The published text does not discuss model risk management or third-party risk, topics that other commentary on AI governance often covers, and it does not set out the report's detailed recommendations. Readers who want those must read the FSB consultation report itself. The remarks also do not announce any change to U.S. regulation.

What it means

For banks, insurers and other institutions, the practical takeaways are modest but clear. First, document how and where AI is used, and assess materiality. Second, scale controls to risk rather than applying one standard everywhere. Third, consider responding to the consultation if the proportionality balance matters to the institution. For customers, the remarks do not change any rights and they do not announce any new rule. This article is general information and not legal advice.