Scenario description
Assume that by 2027-03-31, at least one major U.S. banking regulator—the Federal Reserve, OCC, or FDIC—or a UK banking regulator such as the Bank of England or Prudential Regulation Authority imposes materially binding supervisory expectations, remediation requirements, consent-order obligations, model-risk conditions, deployment limits, auditability requirements, or shutdown-control requirements on banks’ use of agentic AI, autonomous cyberdefense agents, Chinese-developed AI models, or other AI-linked models. The requirement applies to one or more large banks or to a supervised class of banks in the United States or United Kingdom. Exact compliance spending, model inventories, covered vendors, technical standards, and exemptions are not specified unless published by the regulator.