Big US banks block outside AI agents from customer accounts

Several large US banks said in the week of October 12, 2026, that they will not let outside AI agents log into customer accounts or move money. JPMorgan Chase, Bank of America and Citi all set limits, according to reports by CNBC and the Financial Times.

Citi said it does not allow third-party AI agents or assistants to log into a customer's account or make transactions on its digital banking platforms. The bank said the goal is to protect customer accounts and information. Customers can still give limited, read-only access to their account data through certain open-banking providers.

At JPMorgan, CEO Jamie Dimon warned about "rogue agents" during the bank's earnings call on October 13. A spokeswoman told CNBC that third parties cannot use customers' banking login details to act for them. Outside agents can still see some data, such as balances, through connectors like Plaid, but cannot send money or pay bills. Bank of America is limiting access while it decides who is responsible for payments made by AI. CEO Brian Moynihan told CNBC that a human should approve money leaving an account.

The move follows the launch of Meta's Muse personal AI assistant in September. According to FStech and PYMNTS, banks are worried about fraud and unclear responsibility when an agent makes a mistake. Some reports also say banks fear agents could move idle cash to accounts that pay higher interest. The Financial Times reported that Wall Street executives contacted the Federal Reserve about the issue. The Fed declined to comment.

It is not yet clear when, or under what rules, banks might allow outside agents to act on accounts. JPMorgan gave no timeline. Brokerage Charles Schwab said it plans controlled, read-only connections for outside AI tools.