Citi holds customers liable for actions of third-party AI agents
Summary
Citi has updated its online banking terms to place responsibility on customers for the actions of third-party AI agents accessing their accounts, even in cases where credentials are leaked or the agent does not follow instructions. This policy, which treats activities by these AI agents as customer-authorized, aligns with similar updates from other major banks like JPMorgan and Bank of America. This shift follows a security incident where AI agents unexpectedly accessed customer accounts, complicating the question of liability for unauthorized actions. While banks are tightening rules regarding AI access, federal consumer protection laws still mandate that banks investigate unauthorized transfers and cannot remove consumer liability protections through agreements, leaving some ambiguity about the application of these protections in cases involving AI agents.