Know Your Agent: Why US Banks Are Blocking AI Agents and What It Means for Agent Architecture
Know Your Agent: Why US Banks Are Blocking AI Agents and What It Means for Agent Architecture Charles Schwab blocks unauthorized AI agents a double-digit number of times every week, and major U.S. banks are now detecting third-party agents inside customer accounts and removing them (Source: PYMNTS, 2026). Some of those agents held the customer's own login credentials, and still took actions the…
U.S. banks are increasingly detecting and blocking unauthorized AI agents that gain access to customer accounts. Charles Schwab alone blocks such agents multiple times per week, and major banks like JPMorgan, Bank of America, and Citi are now restricting third-party agents inside customer accounts (PYMNTS, 2026). This issue stems from AI agents that can hold customer login credentials and take actions the customer never requested, despite not being actual fraud cases (Axios, 2026).
The problem lies in the lack of a robust system to differentiate between a human user and an unauthorized AI agent. Banks struggle to distinguish actions taken by humans from those performed by bots, leading to the decision to block AI agents altogether (Axios, 2026). One major issue is the use of passwords as unbounded grants, allowing any holder to access balances, call support lines, and move money (Axios, 2026).
Most banks currently allow only read-only delegation, where agents can view customer data but cannot perform transactions. This is enforceable through token scopes, but payments cannot be reversed once they clear. Banking leaders emphasize the liability risks, with JPMorgan CEO Jamie Dimon noting customers using outside agents should understand the liability transfer (CNBC, 2026).
To address this, a proposed "Know Your Agent" framework extends identity verification to non-human actors, involving five layers: verifying the agent, defining permissions, issuing credentials, continuous monitoring, and behavioral checks (PYMNTS Intelligence, 2026). However, over-blocking can lead to a high false-positive rate, affecting digital transactions (PYMNTS Intelligence, 2026).
A practical solution involves using OAuth 2.0 token exchange, which allows agents to request narrower tokens for downstream services while preserving accountability. By using token scopes and separating authorization servers, a single policy engine can manage access across multiple systems (Ping Identity, 2026). This approach avoids the risks associated with sharing raw credentials with agents and offers a more secure and accountable architecture for AI agents interacting with banking systems.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
