Federal Reserve Vice Chair for Supervision Michelle Bowman on Tuesday called for financial regulators to adopt a more proportional approach to artificial intelligence oversight, particularly for smaller institutions. Speaking at a virtual event hosted by the Financial Stability Board, Bowman argued that rules designed for large, complex AI deployments should not be applied uniformly across the industry.
Bowman's remarks coincided with the release of the FSB's consultation report, Sound Practices for Responsible Adoption of Artificial Intelligence, a joint effort by the FSB, the Federal Reserve, the Treasury, and the Securities and Exchange Commission. The document is expected to be finalized later this year as part of the U.S. presidency's G-20 agenda.
Bowman stressed that her views were personal and did not necessarily reflect those of her Federal Reserve Board colleagues or the Federal Open Market Committee. She emphasized that "what works or is a consideration for larger institutions using AI in complex applications is not appropriate for smaller institutions with less complex AI uses." The FSB document itself supports this framing, noting that more robust practices may suit large, complex, and highly interconnected institutions, while smaller or less complex firms may only need a modified subset of the sound practices.
The vice chair argued that regulatory attention should track actual risk rather than the mere presence of AI. Institutions should be clear about how they use AI tools and whether that use is material to their operations or regulatory obligations, as materiality should shape governance and control expectations. Lower-risk applications, she said, warrant a lighter supervisory touch.
The FSB document is not limited to banks; it applies across all types of financial institutions, including asset managers, broker-dealers, and wealth platforms. One consultation question specifically seeks additional case studies from nonbanks. The document highlights AI use in customer profiling, investment planning, portfolio management, and rebalancing, and identifies mis-selling and unsuitable recommendations as consumer-protection risks regulators are monitoring.
Bowman noted that the Fed has tracked bank use of AI for nearly a decade and has observed steady growth in adoption across institutions of every size, spanning a wide range of use cases. She said the central bank's supervisory approach has aimed to support firms that want to adopt the technology responsibly, and that this experience has fed directly into the FSB's global report.
The consultation report includes case studies illustrating governance and control practices that could apply in comparable situations, though Bowman stressed these examples are not the only acceptable route to responsible AI adoption. The document is organized around 12 sound practices covering governance and the full AI lifecycle, with the practice on materiality and risk assessment most directly aligned with Bowman's proportionality argument.
Bowman invited feedback on where the draft practices might be too rigid or fail to account for differences in size, complexity, and risk across institutions. She also asked stakeholders to flag any material risks the report may have overlooked or areas where firms would benefit from additional clarity. The FSB is accepting responses to the consultation until July 22, 2026, via an online submission form and intends to publish responses on its website unless respondents request otherwise.
For wealth managers, the scope of the FSB document brings AI governance directly into their operations. As noted in a recent InvestLin article on Robinhood opening its trading platform to third-party AI agents, the oversight questions are becoming more pressing. Similarly, the FINRA expansion of supervision exam priorities for 2026 underscores the regulatory focus on technology-driven risks.
