Federal Reserve scrutiny of artificial intelligence use at financial companies is increasing, with the OCC and FDIC also pressing banks on AI controls, vendor risk, data access and governance. The shift gives CFOs and finance directors a clear signal that AI adoption in banking is moving from innovation budget to supervisory evidence.
The focus is not limited to customer-service chatbots or back-office productivity tools. Financial companies are using AI across lending, fraud detection, customer verification, regulatory monitoring, sanctions screening and risk management. That creates efficiency gains, but it also places model governance, explainability, cyber controls and operational resilience under greater pressure. For banks, the question is no longer whether AI can reduce cost or improve decision-making; it is whether the institution can prove each system is controlled, monitored and accountable.
The OCC has already issued revised model-risk guidance with the Federal Reserve and FDIC, while stating that generative AI and agentic AI sit outside that framework. Federal Reserve Vice Chair for Supervision Michelle Bowman has also linked AI supervision to the Financial Stability Oversight Council’s wider work on cybersecurity and risk management. That distinction matters for finance teams because many AI tools now sit between model risk, third-party risk, cybersecurity and conduct regulation, rather than inside one neat compliance box.
The U.S. Treasury and FinCEN remain part of the wider risk context because banks using AI in anti-money laundering, customer screening and suspicious activity workflows still need defensible controls. Anthropic and its Mythos model have drawn attention across the sector as financial institutions examine how powerful third-party AI systems could interact with legacy banking infrastructure. For JPMorgan Chase, Bank of America, Wells Fargo and Goldman Sachs, the practical burden is likely to fall on governance committees, risk teams and technology procurement, not only innovation labs.
For CFOs and finance directors, the cost implication is immediate. AI projects will need stronger documentation, clearer ownership, vendor due diligence, audit trails, access controls, human intervention protocols and contingency planning. Budgeting for AI in financial services can no longer be treated as software spend alone; it now includes legal, compliance, cyber, assurance and operational-risk costs.
The broader financial sector should expect regulators to keep using existing supervisory powers before writing detailed AI-specific rules. Finance teams that can map where AI is used, who owns each system, which vendors are involved and how failures are contained will be better placed when examiners ask for evidence. Institutions without that control map may find that AI savings are quickly offset by governance, remediation and regulatory-response costs.
More From Finance Monthly:
Federal Reserve Shift Fuels Fears as Borrowing Decisions Get Harder
JPMorgan Tries to Offload $4 Billion in Private Equity Debt as Cracks Spread Across Wall Street
