US bank accounts at risk as unauthorised AI agents gain access

Unvetted third party AI agents have raised alarm as they gain access to major US banks while showing non-permitted and unpredictable behaviour like operating inside customer accounts.

As reported by Axios citing people privy to the matter, some of unauthorized agentic AI also attempted to take actions using valid login credentials provided by users.

Banks have detected AI agents contacting call centers and attempting to take actions independent of explicit user instructions.

Executives fear agents with login access could independently transfer funds or make unauthorized purchases.

Speaking about the banks, the major ones including Bank of America, JP Morgan, Wells Fargo, Morgan Stanley are taking effective measures to contain these risks posed by these agents, as per the report.

Financial institutions are also racing against the clock to update their terms of service while blocking unauthorized permission and improving platform security.

Bank representatives including Chase advised the users to not give their usernames and passwords to any type of AI tool or third parties. Moreover, Morgan Stanley for now ruled out the instance of rogue AI agents gaining access to platforms.

The rapid advancement of AI technology is revealing security vulnerabilities faster than banks can address them, rendering even substantial financial investments insufficient to completely solve the issue.

The banks acknowledge the gap between AI integration and better security. To balance this, most banks are pivoting toward permitting read-0nly access which allows AI agents to view financial data without taking part in transactions.