Major US banks detect unauthorized third-party AI agents in customer accounts

Banking systems are now having to deal with AI tools that operate through customer logins, a setup that sits on top of existing account access rules at firms such as Morgan Stanley, JPMorgan, Bank of America, and Citi. The concern is not a new product launch but an unauthorized layer of automation entering accounts that were built for human use. That puts pressure on how banks define who is acting on behalf of a customer and who is not.

For people who use financial apps and connected services, tighter restrictions on third-party agents can limit what those tools are able to do once they are inside an account. Banks may also have to redraw responsibility when an agent acts with saved credentials rather than a person typing directly into a platform. The practical result is more control at the bank level and less freedom for outside software to sit between customers and their money. (per cryptobriefing.com)

This case shows the pull between automation and account security in a sector where access is tightly regulated and trust is central. If third-party agents can operate through customer credentials, the boundary between helpful delegation and unauthorized access becomes hard to police. Major banks moving together also suggests the issue is not isolated to one platform but touches a broader design choice in digital finance.