Major US Banks Shift Liability for AI Agent Transactions to Customers

Citibank updated its terms of service this week to inform customers that they are responsible for any activities conducted through third-party AI agents, even in cases of hacking. The policy change specifies that if an AI tool leaks credentials or is compromised, the customer remains liable for any subsequent unauthorized account activity. This follows similar policy announcements from JPMorgan and Bank of America last month, as well as a recent update from PNC and a warning sent to customers by Citizens.

AI agents are increasingly being granted credentials to bank accounts and financial portals to analyze spending, offer investment advice, and manage transactions or bill payments. While AI startups view the personal finance sector as a significant revenue opportunity, financial institutions have expressed concerns regarding the unreliability of these automated actors. Citi's updated Online User Agreement aims to provide clarity regarding the risks of sharing banking credentials with third-party tools.

“We recently updated our Online User Agreement to provide greater clarity about those risks and to reinforce that customers who choose to grant account access to third-party AI agents are responsible for activities conducted through those agents,” a Citi spokesperson told TechCrunch.

The shift in liability reflects broader tensions between AI developers and service providers. While AI labs suggest that agents will revolutionize consumer engagement with service providers, banks appear intent on controlling the interaction pathway. Citi noted that customers may use AI agents provided directly by the bank without facing the same additional liability associated with third-party tools.

Industry responses to the rise of AI agents vary by capability. Anthropic, which launched a personal finance tool earlier this month, has positioned its product as a "read-only" service that cannot move, manage, or invest money. This stands in contrast to the broader trend of agents capable of executing transactions, which has prompted banks to distance themselves from the financial consequences of automated errors or security breaches.

FI Journal
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.