Third-party AI agents are knocking on the doors of customer bank accounts just as regulators begin using AI to scrutinise the banks themselves. That leaves lenders with a hard problem.

According to Corlytics, they must separate legitimate customer activity from unauthorised actions, and they must be able to show how their controls, decisions and lines of accountability fit together.

Picture the scenario. An AI agent attempts to access a customer’s account and the bank blocks it. That single decision raises questions about authorisation, fraud controls and what the customer actually intended. Was the agent a rogue actor, or was it simply carrying out the customer’s instructions?

According to Axios, several major US banks have detected third-party AI agents accessing customer accounts, with some reportedly going beyond what users had asked them to do. A number of these banks are now limiting access while they work out how to distinguish genuine customer activity from behaviour that should be stopped.

Regulators are adding another layer of scrutiny. The Financial Times reported that the FCA intends to use AI to automate more of its supervisory and enforcement work. The watchdog already applies AI to documents, emails and interviews, and it wants to extend this across authorisations, supervision and enforcement.

”, the question is who is really watching whom, and who holds control. The customer’s agent looks into the bank, the bank monitors the agent, and the regulator examines the bank. Throughout, the bank is expected to know what happened, explain why and identify who was responsible.

Source link