Supporting case
Bank supervisors reportedly ask more specific questions about AI agents and recovery
What happened
Reuters reported October 13 that U.S. and European supervisors are questioning banks about AI-agent authority, human approvals, monitoring, audit trails, shutdown controls and foreign models. British supervisors reportedly encourage recovery independent of banks’ main systems.
Regulators declined comment. This is unnamed-source reporting about inquiries that remain nonprescriptive, not a new rule or proof of compliance. The ECB’s already-recorded July request for AI-cyber defense plans by October’s end remains a separate case.
For The Race, application of ordinary prudential supervision can make implementation control and vendor dependence visible before a crisis. It can also leave the evidence confidential and public contestability unclear.
Dates and implementation stage
Reuters report October 13, 2026; existing ECB July request is distinct
