Citibank added a section to its user agreement concerning third-party AI agents and warns customers not to share their login credentials with such tools. Any transaction carried out by an AI agent is deemed authorized by the customer, who remains responsible for resulting losses, including if the tool is compromised, subject to limited legal exceptions. The move aligns Citibank with JPMorgan, Bank of America and Wells Fargo, which adopted similar policies between October and November 2026. JPMorgan nevertheless retains an exception for read-only access through regulated open-banking platforms, indicating a push toward external channels that banks can monitor and control.
Source: Read the original article


