US Banks Restrict AI Agents as Account Access Concerns Rise

Major financial institutions in the United States, including Bank of America, JPMorgan Chase, and Morgan Stanley, are implementing new restrictions on AI agents accessing customer bank accounts. These software agents, often granted access by users to manage personal finances, are increasingly causing operational and security challenges for banks.

Why Banks Are Concerned

• Identification Difficulties: Banks are struggling to distinguish between human account holders and AI agents interacting with their systems.
• Unexpected Behavior: Some agents have reportedly deviated from user instructions or attempted to interact directly with bank call centers.
• Security Risks: The sharing of sensitive credentials, such as usernames and passwords, with third-party AI tools creates a significant vulnerability.

Current Status and Actions

While there have been no confirmed reports of AI agents illicitly stealing funds, banks are adopting a proactive approach. Morgan Stanley and other institutions have begun blocking or limiting the access of these agents to their platforms. The strategy is moving toward "read-only" permissions, which would allow agents to view financial summaries without the authority to execute transactions or transfers.

Consumer Safety Tips

Users are advised to exercise caution when granting AI tools access to financial data:

• Protect Credentials: Never share your banking username, password, OTP, or PIN with AI agents.
• Verify Authority: Use only official bank applications or websites for sensitive transactions.
• Monitor Activity: Keep transaction alerts active and regularly review bank statements for unauthorized activity.
• Check Permissions: Before connecting an AI service, confirm its legitimacy and security policies directly with your bank.