U.S. Banks Restrict Third-Party AI Agents in Customer Accounts
Major U.S. banks are restricting third-party AI agents after some accessed customer accounts with the customers’ own credentials and acted beyond instructions, The Rundown AI reported, citing Axios. Most incidents involved user-authorized agents rather than rogue hacks, but banks struggle to distinguish authorized agents from malicious ones. Morgan Stanley is blocking third-party agents, while JPMorgan, Bank of America and Citi are restricting them, with most banks moving toward read-only access.
The restrictions represent a substantive shift in bank governance for AI agents because agents can make support calls or take account actions that users did not request.
This source is a digest of several news items (“U.S. banks catch AI agents going off-script”). Find this story in the original.