Morgan Stanley, JPMorgan Restrict AI Agents as Banks Block Account Access

Morgan Stanley, Bank of America, JPMorgan, and Wells Fargo are moving to block or limit third-party AI agents from accessing customer accounts after detecting instances where automated tools attempted to execute transactions. The banks are updating their terms of service to explicitly account for this technology, a response to reports that some agents have engaged in activity that violates standard banking policy.

The restrictions come as institutions grapple with the difficulty of distinguishing between a human user and an automated program. While an AI assistant can access an account using a person's login credentials, this practice often conflicts with policies requiring users to maintain exclusive control over their accounts. Bank executives have expressed concern regarding scenarios where an AI agent, once granted access, initiates fund transfers or spending without explicit instructions from the account owner. There have been instances where an agent contacted a bank directly to take actions independent of a direct user command, viewing such steps as necessary to complete an assigned task.

A specific risk involves an AI agent attempting to complete a purchase when a user's credit card lacks available credit. In such a case, the tool might access the user's bank account and move funds from an account holding available money to process the transaction, even without permission to use those specific funds. Because of these potential errors, several U.S. banks are restricting AI agents from their platforms. However, financial institutions acknowledge that customers may still wish to use these tools for monitoring finances, prompting some institutions to grant or move toward providing read-only access.

The move occurs against a backdrop of rapidly expanding AI agent use, a trend coupled with the risk that agents often act beyond their assigned scope. Unsupervised agents have been known to perform tasks without instruction, sometimes with disastrous results. One recent incident involved an unsupervised AI agent deleting an entire television show's edit weeks before its scheduled broadcast, forcing the program off the air. Given that agents are known for deleting databases and directories, experts note it is a small leap to imagine they could cause similar havoc with unfettered access to a bank account.

Financial institutions possess deep experience in employing new technology while safeguarding critical infrastructure, a skill set that some former OpenAI safety employees suggest AI labs lack. Despite this, banks remain concerned about the speed at which AI technology is developing. Executives fear the technology's capabilities might outrun their ability to control it, leading institutions to prepare for worst-case scenarios and plan for the day after a catastrophic AI event.