Major US banks crack down on AI agent use by customers — telling them they’re on their own if it gets hacked or makes a mistake

A photo of a Citibank sign in New York City next to an image of a digitized robotic face
(Image credit: Shutterstock)

  • Major US banks have changed their terms of service regarding AI agents, telling customers they are at fault if something goes wrong
  • If a customer has given an AI agents access to their account, the banks will treat any action taken by the agent as authorized by the customer — even if its a mistake
  • AI companies are introducing "read only" tools to allow users to analyze financials without giving agents purchasing power

Citibank has followed in the footsteps of JPMorgan and Bank of America in changing their terms of service this week to leave customers fully at fault if an AI agent with access to their bank account makes a mistake — or even if it gets hacked.

Banking giants Citizens and PNC have mirrored Citibank’s changes, refusing to accept responsibility for the actions of third-party AI agents given access to accounts by customers.

It was only a matter of time really, given how many people not only consult their AI before making financial decisions, but actually let their AI assistants run rampant with their bank details to analyse their spending and make purchases on their behalf.

Latest Videos FromTechRadar

Credit card

(Image credit: Michal Jarmoluk from Pixabay)

You’re on your own, Citibank says

“As third-party AI tools become more widely available, we want customers to understand the risks associated with sharing banking credentials or authorizing those tools to access their accounts,” a Citibank spokesperson said (via TechCrunch).

“We recently updated our Online User Agreement to provide greater clarity about those risks and to reinforce that customers who choose to grant account access to third-party AI agents are responsible for activities conducted through those agents.”

The issue stems from violations of the number one cybersecurity rule: don’t give anyone your login details. For some people, that apparently doesn’t include an AI agent. Typically a bank’s terms of service require a customer to log in and perform any financial transactions themselves, and AI agents have introduced a new liability that is tricky to navigate — until now.

Basically, if a customer gives a third-party AI agent permission to access a bank account, the bank treats it as an authorized action by the customer. The ‘third-party’ part of that statement is important because Citibank has deployed a range of AI tools for customers to use. If something goes wrong while using one of those, customers won’t face any extra liability.

Given the level of risk involved in providing an AI agent full access to your finances, some AI giants have begun pulling back on the levels of access customers can provide. Gemini for example has introduced a “read only” feature via Google Wallet for analysing spending and tracking payments. Anthropic launched a similar tool earlier in November.

The problem with giving an AI agent access to your finances isn't just that it might make a mistake. It's that, increasingly, your bank is making it very clear that you'll be the one paying for it.


Google logo on a black background next to text reading 'Click to follow TechRadar'

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.


An Apple MacBook Air against a white background
The best laptops for all budgets

➡️ Read our full guide to the best laptops
1. Best overall:
Apple MacBook Air 13-inch M5
2. Best budget:
Apple MacBook Neo
3. Best Windows 11 laptop
Microsoft Surface Laptop 13-inch
4. Best thin and light:
Lenovo Yoga Slim 9i
5. Best Ultrabook
Asus Zenbook S 16


TOPICS
Benedict Collins
Senior Writer, Security

Benedict is a Senior Security Writer at TechRadar Pro, where he has specialized in covering the intersection of geopolitics, cyber-warfare, and business security.

Benedict provides detailed analysis on state-sponsored threat actors, APT groups, and the protection of critical national infrastructure, with his reporting bridging the gap between technical threat intelligence and B2B security strategy.

Benedict holds an MA (Distinction) in Security, Intelligence, and Diplomacy from the University of Buckingham Centre for Security and Intelligence Studies (BUCSIS), with his specialization providing him with a robust academic framework for deconstructing complex international conflicts and intelligence operations, and the ability to translate intricate security data into actionable insights.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.