Category: Industry Analysis | Reading time: 3 min
Axios reported on October 22 that major U.S. banks, including Morgan Stanley, Bank of America, JPMorgan and Wells Fargo, are detecting third-party AI agents operating inside customer accounts. According to the report, some of those agents have attempted actions that appear independent of what the customer asked them to do.
How the agents get in
In most cases the agents are not breaking in. They log in with the customer's own credentials, with the customer's permission. That still runs against most banks' guidelines, and it leaves the bank seeing what looks like a normal customer session. Banks cannot easily tell whether a person or a piece of software is on the other end.
What banks are doing
Axios reports that banks are responding by blocking agents and rewriting their terms of service. The report is based on unnamed sources, and the banks named have not publicly detailed their responses, so the specifics should be read as reported rather than confirmed.
Why it matters
Shared credentials collapse two very different things into one identity: the customer, and whatever software the customer handed their password to. When an agent takes an action the customer did not intend, neither the bank nor the customer has a clean record of who authorized what. For regulated institutions, the questions become practical: which agents are allowed to act, under whose authority, within what limits, and what evidence exists when an agent's action is disputed. Blocking buys time, but customers are already using these tools, and banks will need a way to recognize agents as agents and decide on each action before it happens.
Source: Axios