Skip to content

Citi: Customers Liable for Hacked Rogue AI

Citi revised its online banking terms in November to treat actions taken with credentials given to a third-party AI agent as customer-authorized, even if the agent ignores instructions or leaks the credentials. Citi says it need not check whether an agent followed the customer’s instructions, subject to applicable law. The provision does not apply to Citi’s own AI agents.

Federal rules still require banks to investigate genuinely unauthorized electronic transfers and limit customers’ liability for fraud; a banking agreement cannot waive those protections. Whether they cover a particular loss involving a customer-authorized AI agent depends on how the transaction occurred. PNC also updated its agreement, allowing certain bank-approved connections but barring customers from giving third-party AI agents passwords or one-time codes.

Source

Share