US banks block outside AI agents in customer accounts

Major US banks are detecting third-party AI agents operating in customer accounts, including cases using customer-provided credentials, and are already restricting such access. The conflict extends beyond banking: any site where customers take meaningful action must decide whether an agent is a valid user, a bot, or a security incident.

• Banks struggle to tell human-directed activity from agent-directed activity
• Customers believe access is authorized; bank policies forbid sharing credentials
• Liability for agent errors has no clear governance model
• Enterprises will rewrite policies and authentication over the next 2–3 years