Technology news and discussion on TalkFuse.
Citi makes customers liable for hacked AI tools accessing accounts
Citibank has updated its terms to say customers are responsible for activity by third-party AI tools they authorise to access their accounts, including unauthorised activity if the tool is hacked or leaks their credentials. The policy reflects banks’ concerns about unreliable automated systems and clarifies that they may not reverse transactions made by AI agents.
JPMorgan Chase and Bank of America announced similar policies last month, while PNC changed its terms on the same day as Citi and Citizens warned customers by email. AI tools that analyse spending or make payments could give consumers new ways to manage money, but banks also face security risks and competition over how customers access financial services; Anthropic’s recently launched finance tool is read-only.
In short
- Citi says customers bear responsibility for third-party AI agents’ account activity.
- Similar policies are spreading among major US banks.
- AI finance tools raise both security and competition concerns.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Banks can reasonably argue that customers who choose to give an outside AI tool account access should bear responsibility for the risks that come with that choice, especially if the tool is compromised or misuses credentials. Clear terms may help banks manage fraud and security risks while encouraging developers to build safer systems and consumers to assess permissions carefully.
The case against
Customers can reasonably argue that they should not automatically absorb losses when an authorised tool is hacked or leaks credentials, particularly if the resulting transactions were not genuinely approved by them. Broad liability terms could weaken consumer protection and trust in useful financial tools, while banks may be better placed to set secure access standards and distinguish valid instructions from criminal activity.
Explained simply
Banks hold customers’ money and provide tools for checking balances, paying bills and moving funds. Some newer AI services can connect to bank accounts to analyse spending or make payments, usually after a customer grants access.
Citi, JPMorgan Chase and Bank of America are among the banks setting out how responsibility works when customers authorise outside services to use their accounts. These services are built by companies separate from the banks, so questions can arise over who is responsible if an account is accessed or a payment is made without the customer’s intention.
The issue matters as automated financial tools become more common and banks weigh their usefulness against security risks. The terms governing account access can affect whether a customer can recover money after unauthorised activity.
Read the full article at the source →
Originally published by TechCrunch as “More banks say customers using AI agents are on their own — even if they get hacked”.