Federal Reserve quizzes banks on AI use as stability fears mount

The scrutiny is not arriving in a vacuum. Banking regulators are stepping up efforts to contain growing artificial intelligence risks, and the sources describe this as adding to prior scrutiny rather than a fresh start. What has changed is the recognition that advances in frontier AI may demand costly investments in cyberrecovery, closer attention to dependencies on external technology providers, and, in some jurisdictions, more formal regulatory guidance.

The concern is not only how banks use these models day to day. Some of the six people familiar with the matter pointed to the potential for cyberattacks on banks, or for rogue agents to undermine global financial stability. That framing puts AI oversight closer to the traditional stability mandate regulators already carry, rather than treating it as a purely technical question for bank IT departments.

The practical effect appears to be a two-track approach, based on the accounts given: direct questioning of banks by U.S. regulators, alongside a possible move toward formal guidance in some jurisdictions. How far that guidance goes, and how quickly, is not specified in the accounts.