What's Happening?
Citigroup and PNC Financial Services have updated their account agreements to place the responsibility for losses incurred through third-party AI agents onto their customers. Citigroup's revised terms,
effective November 2026, explicitly state that if customers provide their banking credentials to outside AI agents, any activity conducted with those credentials will be considered the customer's own, even in cases of data leaks. This clause, identified as Section H, specifically targets third-party AI agents and is subject to applicable law. A Citigroup spokesperson clarified that this change is intended to inform customers about the risks associated with sharing login details or granting account access to external tools, emphasizing that customers authorizing outside agents are accountable for their actions. PNC amended its rules on November 18, 2026, but continues to allow customers to share account data via an API built on the Financial Data Exchange standard, which enables app connections without exposing banking passwords. These revisions follow similar actions taken by other major U.S. banks in October, including JPMorgan Chase, Bank of America, Charles Schwab, and Wells Fargo, all of whom have implemented various measures to address the risks associated with AI agent usage.
Why It's Important?
This shift in liability by major U.S. banks like Citigroup and PNC has significant implications for consumer protection and the evolving landscape of financial technology. By making customers responsible for actions taken by third-party AI agents they authorize, banks are attempting to mitigate their own exposure to potential fraud and security breaches. This move could lead to increased caution among consumers regarding the use of financial management apps and services that require direct access to banking credentials. It also highlights a growing tension between technological convenience and user responsibility in the digital age. While banks argue these changes clarify risks, consumer advocates may raise concerns about the fairness of placing the entire burden of security on individual users, especially given the complexities of AI and cybersecurity. The differing approaches among banks—some blocking direct credential logins while others advise pre-authorization or simply state 'at customer's own risk'—could create a fragmented regulatory and consumer experience, potentially leading to confusion and varied levels of protection across financial institutions.
What's Next?
The implementation of these revised terms by Citigroup and PNC, alongside similar policies from other major banks, is likely to prompt further discussion and potential action from consumer protection bodies. The Consumer Financial Protection Bureau (CFPB) has previously indicated that its definition of unauthorized electronic fund transfers covers those made with credentials stolen from third parties, suggesting a potential conflict with the banks' new contractual language. This could lead to legal challenges or regulatory guidance aimed at clarifying the extent of customer liability versus bank responsibility in cases involving AI agents. Consumers will need to become more vigilant about the terms and conditions of their banking agreements and exercise extreme caution when using third-party financial tools that request direct access to their accounts. The financial technology industry may also see a push towards more secure API-based data sharing methods, like the Financial Data Exchange standard, as banks increasingly restrict credential-based access to mitigate risks for both themselves and their customers.
Beyond the Headlines
The broader implications of banks shifting AI agent liability to customers extend beyond immediate financial transactions. This trend underscores a fundamental re-evaluation of trust and responsibility in the digital financial ecosystem. As AI tools become more sophisticated and integrated into daily financial management, the line between authorized and unauthorized activity becomes increasingly blurred. This raises ethical questions about who bears the ultimate responsibility when autonomous systems, even those authorized by the user, cause financial harm. It could also spur innovation in secure authentication methods and data sharing protocols that minimize the need for customers to directly share sensitive credentials. Furthermore, this development highlights the ongoing challenge for regulators to keep pace with rapid technological advancements, ensuring that consumer protections remain robust in an environment where AI agents act as intermediaries in financial transactions. The long-term impact could be a redefinition of consumer rights and bank obligations in the era of AI-driven finance.








