AI agents detected inside customer bank accounts, report says

A new Axios report reveals that major U.S. banks are detecting unauthorized activity from third-party AI agents operating within customers' online accounts. Some customers have granted personal AI agents, which are not vetted by banks, access to their financial accounts. Madison Mills, a senior AI reporter for Axios, explained that in many cases, customers are simply providing their username or password to an unvetted AI agent that goes against the bank's terms of service.

In other instances, Axios found evidence of AI agents taking unauthorized actions in customer accounts, such as making complaints to the bank without the account holder's consent. The White House AI czar, David Sacks, clarified that these activities are likely instances of user-authorized AI agents, which differ from hackers deploying AI for malicious purposes.

Sacks suggested that banks could limit agents' access by granting them read-only permission, allowing them to view spending and bank activity without taking any actions. Banks are taking steps to address the issue, including informing customers about the risks of sharing credentials with AI agents. Some banks, like Chase and JPMorgan Chase, have taken action to make customers aware of these risks, while others, like Bank of America and Wells Fargo, have not yet commented on the matter.