Bank supervisors press lenders on how AI agents are limited, audited and switched off

Banking supervisors are stepping up questions about AI, according to a Reuters report based on six sources, summarised by Resultsense.

In the UK, supervisors including the Bank of England are encouraging the largest lenders to consider “bare metal” recovery systems kept separate from their main technology estate, and to use frontier models to find and fix weaknesses in their own systems. The approach is described as collaborative rather than new requirements.

In the US, the Federal Reserve, the OCC and the FDIC are asking what limits banks place on AI agents, how agent actions are approved, whether decisions can be audited and whether a human stays in the loop — still in a fact-finding phase. In the euro area, the ECB gave lenders until the end of October to produce plans against AI-enabled cyber threats and asked them to map AI use, including third-party suppliers, so systems can be stopped and restarted in a controlled way.

Editor’s note: we could not access the original Reuters report; this summary relies on Resultsense’s account of it.