Axios: Unvetted AI Agents Are Operating Inside US Bank Accounts
Major US banks are detecting third-party AI agents operating inside customer accounts, and in some cases the agents appear to act without a direct instruction from the account holder, sources tell Axios. Morgan Stanley, Bank of America, JPMorgan, Wells Fargo and others are moving to contain the risk.
The usual pattern is a customer handing an assistant their login and password so it can do financial work, which breaks rules requiring the human to log in. Some banks have also tracked agents contacting call centers or attempting actions of their own. Several executives told Axios they cannot reliably tell whether agent activity is user-directed, or whether a given process is an authorized assistant or something that gained access on its own:
- Morgan Stanley has started blocking third-party agents outright.
- JPMorgan, Bank of America and Citi are also restricting agent access.
- Most banks are heading toward read-only access, the same arrangement already given to budgeting apps such as Quicken.
Two of the named banks pushed back on the framing. A Chase spokesperson said customers should not share usernames and passwords with anyone, and a Morgan Stanley spokesperson said the bank has not observed rogue agents accessing its platforms. Banks are preparing for the worst case anyway, with “day after” plans to secure or restore customer access quickly.