Major US banks are taking steps to block unvetted artificial intelligence (AI) agents from accessing customer accounts after reports emerged that third-party software was attempting to carry out transactions without customers’ authorisation.
According to a report by Axios published on October 22, banks detected AI agents attempting to perform actions within customer accounts that had not been explicitly approved. Reuters subsequently reported that major US financial institutions were moving to restrict access by these agents.
AI agents are software tools designed to perform tasks on behalf of users, including managing financial accounts. In some cases, customers reportedly shared their banking login credentials with these tools. However, bank security systems intervened and blocked certain transactions.
The development has raised concerns because transactions initiated using valid customer credentials may appear legitimate to banking systems, even when the specific actions have not been authorised by the account holder.
Several major financial institutions have introduced measures to prevent unverified AI agents from accessing their platforms.
Morgan Stanley has blocked AI agents from accessing accounts on its platform. A spokesperson for the bank said it had not observed instances of unauthorised AI agents accessing its systems. Bank of America and Citigroup have also taken steps to restrict access by unvetted agents, according to the reports.
JPMorgan Chase and Charles Schwab Raise Concerns
JPMorgan Chase and brokerage firm Charles Schwab had addressed the risks associated with AI agents earlier in October.
JPMorgan Chase CEO Jamie Dimon said the bank would not allow external AI agents unrestricted access to its systems, citing risks to customers and the institution.
The growing use of AI tools to automate financial activities has prompted banks to reassess how third-party software interacts with their platforms. While such tools can help users manage accounts and perform routine tasks, unrestricted access may create security and authorisation risks.
Financial institutions are therefore working to ensure that automated tools operate within clearly defined permissions and that sensitive transactions remain subject to appropriate safeguards.
