Wall Street banks restrict AI agents from making automated financial transactions

JPMorgan Chase, Bank of America, and Citigroup are tightening restrictions on third-party artificial intelligence agents accessing customer accounts, Cryptonews.net reported. The measures target unauthorized transactions conducted by autonomous software capable of logging into bank accounts and moving money without human intervention. [1]

The crackdown highlights security challenges for traditional financial institutions as autonomous software operates at high speeds while existing bank security infrastructure relies on human verification methods. Cryptonews.net reported that current banking security systems remain structured around human behavior, manual approvals, and passwords. [1]

JPMorgan Chase is the largest bank in the United States and the world's largest bank by market capitalization, according to Wikipedia. It is categorized as a systemically important financial institution by the Financial Stability Board. [4]

The banking sector restrictions come amid wider deployment of agentic tools across tech platforms. Anthropic, for instance, develops Claude, a series of large language models used to power agentic software tools such as terminal coding agent Claude Code and non-programmer tools such as Claude Cowork. [2]