Skip to content
Regulation & PolicyLatest

Banks Update Terms on AI Agent Actions, Shifting Liability Amid Consumer Protection Uncertainty

Published

Major U.S.

MissedBlock Desk · 4 min read

Banks Update Terms on AI Agent Actions, Shifting Liability Amid Consumer Protection Uncertainty

Major U.S. banks are revising their online banking terms and digital agreements, a move that increasingly places responsibility on customers for actions taken by third-party AI agents accessing their accounts. This shift, observed at institutions including Citi, JPMorgan, Bank of America, Citizens Bank, and PNC, addresses the growing use of AI tools capable of acting on behalf of users. However, the precise application of existing consumer protection laws to these evolving scenarios remains a significant area of uncertainty.

Key Bank Policy Updates and AI Agent Capabilities

Citi revised its online banking terms in November, stipulating that actions performed using credentials provided to an outside AI agent are considered authorized by the customer. This holds true even if the AI agent disregards instructions, exceeds its permissions, or shares those credentials. According to Citi’s updated terms, the bank is not obligated to investigate whether the agent adhered to the customer’s original prompts, subject to applicable law. This clause does not extend to AI agents developed and offered by Citi itself.

JPMorgan and Bank of America announced similar policy adjustments in October. More recently, Citizens Bank informed its customers via email this week that they could be held accountable for the actions of their AI assistants. PNC also modified its digital agreement on Wednesday, with its updated terms differentiating between bank-approved connection methods and unrestricted access. These policy updates follow the detection of AI agents accessing customer accounts by major banks around October 22.

An AI agent differs from a conventional chatbot by its capacity to act autonomously on a user’s behalf, potentially navigating websites, accessing financial records, or initiating payments. Consequently, a mistaken instruction or a compromised login via an AI agent carries greater financial implications than a simple information request.

Consumer Protection Framework and Emerging Gaps

The Office of the Comptroller of the Currency has issued guidance for AI examiners, covering aspects like outside agents, payment controls, and human oversight. However, this guidance is nonbinding and does not impose new requirements on banks.

Existing U.S. federal consumer protection laws generally mandate that banks investigate unauthorized electronic transfers and limit customer liability for fraud. These protections cannot be unilaterally waived through banking agreements. Nevertheless, their precise application to transactions executed by customer-authorized AI agents remains a developing area of uncertainty. The Consumer Financial Protection Bureau has stated that banks cannot use agreements to circumvent these legal protections. Current regulations distinguish between theft by an external party and transfers that were actually authorized by a customer or someone granted access. Therefore, whether a specific AI-related loss is deemed unauthorized hinges on the nature of the transaction’s occurrence.

Navigating the Uncertainties

This evolving landscape presents several uncertainties. The applicability of U.S. federal consumer protection rules to transactions initiated by customer-authorized AI agents is not yet definitively established. How a particular AI-related loss is classified as unauthorized depends on the specifics of how the transaction took place. Furthermore, the extent to which banks are required to investigate AI agent actions versus actions explicitly authorized by the customer is still being clarified.

While banks are updating their terms of service, these changes do not automatically nullify all existing consumer fraud protections. However, the clarity and enforceability of these protections in the context of AI-driven financial transactions are subjects of ongoing discussion and potential future regulatory clarification.

Why This Matters

This situation highlights a critical juncture where technological advancement in AI intersects with established financial regulations. As banks update their terms to address AI agent access, the core issue becomes the potential impact on consumer liability and the existing safeguards designed to protect individuals from financial fraud. The fact that multiple major banks are implementing similar policy shifts underscores the growing importance of this issue for millions of customers.

Broader Context

The move by banks follows a period of increased detection of AI agents accessing customer accounts, with major institutions noting such activity around October 22. The Office of the Comptroller of the Currency has since issued AI examiner guidance, and the Consumer Financial Protection Bureau has commented on the limitations of banking agreements in waiving legal protections. This indicates a developing regulatory environment attempting to keep pace with financial technology.

Banks Update Terms on AI Agent Actions, Shifting Liability Amid Consumer Protection Uncertainty · MissedBlock