A group of large U.S. banks have started to block external AI agents that log in to accounts with a customer’s username and password. The banks named by Axios include Morgan Stanley, JPMorgan, Bank of America, Wells Fargo and Citi. The banks say their policies require account holders to log in themselves, and using an AI to do so violates those rules.
Why banks are pulling the plug
Most of the banks want to allow AI to read account data only, not to perform transactions. They say they can’t reliably tell whether an AI is acting on a user’s direct instruction or acting on its own. The risk of an agent moving money or changing settings without clear user consent is considered too high. OpenAI has launched a separate ChatGPT service for U.S. finance that is limited to read‑only data, but the banks are still blocking agents that log in through a browser or tools like Computer Use and the open‑source OpenClaw.
Why it matters
If you use an AI assistant to manage your bank accounts, you may find it can no longer log in for you. You’ll have to interact with your bank directly or use services that only show you data without performing actions. The banks say the move is to protect you from unauthorized transactions, but it also limits the convenience AI tools can provide.
