What happened
Major US banks, including Citi, Wells Fargo, Bank of America, and JPMorgan Chase, have implemented measures to restrict or block third-party AI agents from accessing customer accounts and executing transactions. Citi explicitly stated it does not permit third-party agents to log in or transact, offering only limited read-only access via approved providers. Wells Fargo updated its online access agreement to clarify that users are solely responsible for supervising agents and that agent-initiated transactions are considered authorized unless explicitly revoked. Bank of America and JPMorgan Chase confirmed they are blocking outside agents from accessing customer account information, with Bank of America CEO Brian Moynihan emphasizing the need to keep a human in the loop for payment authorization. These moves follow a September coalition of global banks warning that agentic technology is outpacing industry standards and protections.
Citi, Wells Fargo, Bank of America, and JPMorgan Chase have taken steps to limit third-party AI agents from accessing their platforms and services. Citi stated it does not permit third-party AI agents to log into customer accounts or transact directly, allowing only limited read-only access through approved open banking providers. The bank reported no instances of agents accessing client accounts.
Wells Fargo updated its online access agreement, set to take effect on November 24, which includes a section on third-party AI agents. The agreement states that users are solely responsible for supervising agents and that transactions by authorized representatives, including agents, are considered authorized unless the user explicitly notifies the bank otherwise. The bank also noted that permitting agent access may impact security procedures and that users accept the associated risks.
Bank of America CEO Brian Moynihan confirmed the bank is restricting AI agents' access to consumer deposits and working with companies to develop protocols. He emphasized the need to keep a human in the loop for payment authorization, stating that when an agent is given credentials, it is the user authorizing the payment. JPMorgan Chase CEO also confirmed the bank is blocking outside agents from accessing customer account information.
These actions follow a September coalition formed by Bank of America, Capital One, and four other global banks to warn that agentic technology is moving faster than industry standards and protections. The coalition highlighted the specific risks posed by AI shopping agents. Experts like Andre Durand of Ping Identity noted a 'window of vulnerability' where systems, banks, and users are vulnerable to the speed of AI weaponization, citing concerns over fraud, scams, and compromised agents.
Source details: usatoday.com ↗
Why it matters
This development marks a significant shift in how financial institutions approach the security of agentic AI, moving from theoretical concern to concrete operational restrictions. As AI agents become more capable of executing complex tasks, including financial transactions, the risk of unauthorized or malicious actions increases. By restricting agent access, banks are prioritizing security and consumer protection, potentially slowing the adoption of fully autonomous financial agents. This also highlights the broader challenge of verifying the identity and authority of AI agents in digital systems, with experts calling for improved authentication methods like passkeys and biometrics to ensure human oversight.
The restrictions reflect growing concern among financial institutions about the security risks of agentic AI. A 2026 study by Palo Alto Networks found that two-thirds of CISOs in financial services and software ranked agentic AI among their top three cybersecurity risks, with over one-third naming it their top concern. This indicates a widespread recognition of the potential threats posed by autonomous AI systems in sensitive environments.
The moves by major banks could set a precedent for how the financial industry handles access, potentially influencing regulatory frameworks and industry standards. By requiring human authorization for transactions, banks are prioritizing consumer protection and accountability, which may slow the adoption of fully autonomous financial agents but enhance trust in digital banking systems.
The situation highlights the technical challenges of verifying the identity and authority of AI agents. Current authentication methods, such as one-time passwords sent to email, are considered insufficient because agents often have access to email. Experts are advocating for more robust methods like passkeys and biometrics to ensure that a human is authenticating and authorizing actions, addressing the core issue of distinguishing between human and agent activity.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
Monitor for further regulatory guidance on authentication and authorization in financial services. Watch for updates from the global bank coalition on developing industry standards for trusted agentic commerce. Observe whether other financial institutions adopt similar restrictions or develop their own AI assistants to control agent access. Track consumer adoption rates of AI agents for financial tasks, as restrictions may influence user behavior and trust.
Regulatory bodies may issue new guidelines or regulations regarding the use of AI agents in financial transactions, particularly focusing on authentication, authorization, and liability. The actions of major banks could regulators to formalize standards for trusted agentic commerce.
The global bank coalition, which includes Bank of America, Capital One, and others, may release further recommendations or standards for managing risks. Their efforts to develop protocols for agent access could influence industry-wide practices and potentially lead to new security frameworks.
Financial institutions may develop or expand their own AI assistants, like Charles Schwab's 'Charley,' to provide controlled access to services while blocking external agents. This trend could lead to a more fragmented landscape where banks offer proprietary AI tools rather than relying on third-party agents.
Consumer behavior and trust in AI agents for financial tasks may be affected by these restrictions. As banks tighten controls, users may become more cautious about delegating financial actions to AI, potentially impacting the growth of agentic commerce and the adoption of AI-powered financial services.