Citi Bank is the third-largest commercial bank in the United States. This week, the bank revised its service terms to inform customers who use third-party AI tools to manage their personal finances that any issues that arise will be the sole responsibility of the customers themselves—even if their accounts are compromised.
The scope of AI agents ranges from cutting-edge laboratory models to specially tailored personal finance tools. They are capable of obtaining access to your bank account and other online financial portal credentials, with the aim of analyzing consumption and investment decisions, providing recommendations, and even handling transactions and bill payments.
However, financial institutions are not welcoming these less reliable AI entities into their systems and inform customers that they will not assume any responsibility for canceling transactions initiated, either intentionally or unintentionally, by automated assistants.
A Citibank spokesperson told TechCrunch: "As third-party AI tools become more and more widespread, we want our customers to be aware of the risks associated with sharing bank credentials or authorizing these tools to access accounts. We have recently updated our online user agreement to clarify these risks further and emphasize that customers who choose to grant account access rights to third-party AI agents are responsible for the activities conducted by those agents."
Last month, two major U.S. banks, JPMorgan Chase and Bank of America, announced similar policies. PNC – the eighth largest bank in the United States – also revised its terms today.
Citi's update not only indicates that customers are responsible for whatever their agents may do, but also informs them that if the AI tool leaks credentials or is hacked, customers will also be responsible for whatever the hackers do using their accounts. A large American bank, Citizens, also sent emails to customers this week warning them that they are equally responsible for any actions taken by AI assistants. TechCrunch saw one such email.
As AI startups seek additional revenue, entering the personal finance sector presents an attractive opportunity for businesses. It not only allows them to add value to their customers but also to reap a portion of the fee stream generated by this industry. However, if consumers are already wary about sharing their email or chat records with AI agents, then bank and brokerage accounts pose new trust challenges.
The approach of Anthropic is quite illustrative: the personal finance tool they launched earlier this month is clearly labeled as "read-only," and it "cannot move, manage, or invest your money."
In addition to concerns about cybersecurity, banks also see efforts to enter their own business territories as a competitive threat. The market was briefly shaken by the possibility that AI tools could enable consumers to easily avoid unnecessary fees or low-interest deposit accounts, which would increase banks' profit margins. However, if AI agents become an important way for people to make financial decisions, banks will strive to control this path. Citibank's terms state that the AI agents it provides can be used by customers without having to worry about additional responsibilities.
The reaction of banks is also part of a broader trend: businesses in all industries are facing the same situation – AI laboratories promise that their agents will completely change the way people interact with various service providers, while service providers say “wait a minute.” We have seen this in areas ranging from ride-hailing to catering services, and how these relationships evolve in the end will help determine which AI investments will be profitable.












