US banks clamp down on third-party AI agents amid security fears
Major U.S. banks are blocking external AI agents from accessing their systems because of worries that unverified bots could misuse customer credentials.
A group of leading U.S. banks, including Morgan Stanley, JPMorgan Chase, Bank of America, Wells Fargo and Citigroup, are tightening controls on third-party artificial-intelligence agents that attempt to connect to their services. The move follows incidents where AI bots accessed customer accounts using login details and appeared to perform actions independently of the account holder. While banks already allow certain third-party apps limited read-only access for budgeting, they are now wary of agents that can initiate purchases or transfers.
Some institutions are moving toward permitting only data-retrieval functions for external AI, to prevent unauthorized transactions. A Chase spokesperson warned customers not to share credentials, and a Morgan Stanley representative said no rogue AI incidents have been observed on their platform. Executives acknowledge that rapidly evolving AI tools make it harder to verify whether activity originates from a person, an authorized bot, or a rogue agent.
Why it matters
Uncontrolled AI access could expose millions of bank customers to fraud or unauthorized financial moves.
In this story
