TechCrunch reported on 18 November that Citi changed its online user agreement this week. A Citi spokesperson said customers who give third-party AI agents account access are responsible for what those agents do. TechCrunch says the terms also make customers liable if the tool leaks their login or is hacked. JPMorgan and Bank of America announced similar policies last month. PNC changed its terms on 18 November, and Citizens emailed customers this week.
TechCrunch says agents can be given a customer's bank login to study spending, give advice and pay bills. Anthropic's finance tool, launched this month, is billed as read-only and unable to move money. Citi's terms let customers use agents that Citi provides without extra liability. The report does not say what PNC's change covers, how many customers use third-party agents, or describe any customer loss.
These terms put the risk on whoever hands over a login. At Citi, per TechCrunch, that includes a hack of the agent. TechCrunch does not say whether the other banks cover vendor breaches. Buyers should treat an agent that holds a login as a vendor that holds a credential, and ask whether it can only read or can also pay.