Community Trust ScoreLikely Real
Citi has revised its online banking terms to treat activity conducted by third-party AI agents as actions authorized by the customer. The November update shifts responsibility for unauthorized transfers and credential leaks to the account holder, even if the artificial intelligence tool ignores specific instructions or the customer’s login details are compromised.
More coverage is filed under Finance.
Shifting Liability to Customers
Under the new agreement, any action taken using credentials supplied to an outside agent is considered authorized by the customer. Citi states it has no obligation to investigate whether an AI agent followed the customer’s original prompts, subject to applicable law. The provision explicitly excludes AI agents offered by Citi itself, limiting the scope of the liability shift to external tools.
A Citi spokesperson told TechCrunch that the November revision clarifies the risks associated with sharing banking credentials or granting outside tools access to accounts. This approach aligns with a broader trend among major U.S. banks to tighten policies in response to emerging security threats involving autonomous software.
Industry-Wide Policy Adjustments
Other financial institutions have adopted similar stances. JPMorgan and Bank of America announced comparable policies in October. Citizens Bank warned customers via email this week that they could be held responsible for the actions of their AI assistants. PNC changed its digital agreement on Wednesday, distinguishing between bank-approved connections that provide limited access and scenarios where an agent is given unrestricted access to an account.
PNC’s updated rules prohibit customers from providing third-party AI agents with passwords or one-time security codes. Under PNC’s framework, agents generally cannot initiate payments unless the bank expressly permits it. These changes follow a specific security incident that highlighted the challenges banks face in managing third-party access.
Regulatory and Legal Context
The policy shifts occurred against the backdrop of a security scare detected on October 22, when major banks identified AI agents logging into customer accounts. Executives stated they struggled to determine which AI agents had been properly authorized during the incident. AI agents are capable of acting on a user’s behalf, including navigating websites, accessing financial records, or initiating payments.
The Office of the Comptroller of the Currency has issued AI examiner guidance addressing outside agents, payment controls, and human approval. This framework is nonbinding and does not create new requirements for banks. However, under U.S. federal consumer protection rules, banks generally must investigate unauthorized electronic transfers, and federal rules limit customers’ financial liability for fraud.
The Consumer Financial Protection Bureau states that banks cannot use agreements to waive legal protections. While federal protections generally apply to unauthorized transactions, the application of these safeguards to transactions performed by customer-authorized AI agents remains less clear. The legal landscape continues to evolve as banks attempt to balance the convenience of AI tools with the need for security and regulatory compliance.
