Three major U.S. banking regulators are questioning lenders about how they use and control autonomous AI agents, as concerns grow that poorly supervised systems could expose financial institutions to cybersecurity threats and operational failures, Reuters reported.
The Federal Reserve, Office of the Comptroller of the Currency and Federal Deposit Insurance Corp. are asking banks about the limits placed on AI agents, how their actions are approved and monitored, and whether decisions can be traced and audited, according to people familiar with the inquiries.
Regulators also want to know whether humans remain involved in important decisions, what prevents agents from acting beyond their authority and whether banks can shut down autonomous systems.The agencies declined to comment to Reuters.
The inquiries cover large, midsize and smaller banks and include questions about foreign-developed AI models and agents used in cyberdefense. Regulators are gathering information rather than imposing new prescriptive requirements, the report said.
Agentic AI systems can take multistep actions with limited human intervention, offering banks potential efficiency gains while creating risks if agents misuse tools, act on incorrect information or exceed their permissions.
“The adoption of AI is often outpacing the ability to put the controls in place,” Michael Woodbridge, a London-based cybersecurity partner at Grant Thornton, told Reuters.