Axios reports that JPMorgan, Bank of America, Wells Fargo and Morgan Stanley are blocking or limiting third-party AI agents that sign in with a customer’s own credentials. Many of the limits point toward read-only access.
What the banks are changing
The problem Axios describes is attribution. A bank cannot easily tell a click the customer meant from an agent that went further than the customer asked.
Read-only access, where an agent can look but not move money or change settings, is the direction several of these banks are reported to be taking. It is not a finished industry rule.
Anyone who has connected a chatbot to email, a password manager, or a bank. An agent that can read a balance is a different risk from one that can pay a bill.