Major US financial institutions, including JPMorgan, Bank of America, Citi, and Morgan Stanley, are actively blocking AI agents that attempt to access customer accounts using provided credentials. These banks state that such automated access violates internal policies, which mandate that account holders must log in personally to manage their finances.
The institutions are concerned about security risks, noting that they often cannot distinguish between an agent acting on a user's behalf and one operating independently. While Morgan Stanley has implemented a complete block on AI agents, it reports that it has not identified any rogue AI activity on its platforms so far.
Banks are currently exploring ways to limit AI access to read-only functions, allowing agents to view data without executing transactions. However, technical enforcement remains challenging, as users can grant access through browser-based tools like Computer Use or open-source software such as OpenClaw.
In response to the demand for AI-driven financial management, OpenAI has introduced a dedicated ChatGPT service in the US. This specific tool is designed to provide read-only access to financial data, aligning with the security requirements currently preferred by the banking sector.
