Last Updated on November 22, 2026 by Jade Artry

PERSONAL SAFETY

Banks Are Changing Their AI Agent Rules: What It Means for Your Money

Jade Artry

Jade Artry

Following a series of major security incidents affecting banks, airlines and other services, several US banks have introduced terms that now place the responsibility on customers for certain actions taken by AI agents they allow into their accounts. Citi’s agreement goes further, stating that the bank will treat activity using login details leaked by an agent as authorised by the customer. Here’s what the banks’ terms say, what protections still apply and how to protect your money.

A customer and bank adviser review account documents together

At a glance

What Are AI Agents Doing With Bank Accounts?

An AI agent is a tool that can carry out tasks on your behalf. Depending on the service and the access you give it, an agent could gather information, compare products, manage a schedule or complete transactions.

A financial agent might be asked to check a balance, organise bills or make a payment. To do that, it needs a way to access the relevant account or service.

That access creates an important question: who is responsible if the agent makes a mistake, takes an action you didn’t intend or exposes your information?

Several banks now address that question in their agreements. Their wording places responsibility for certain actions on customers who allow third-party agents to act for them.

What Does Citi Say?

Citi’s online user agreement, updated in November 2026, warns customers against using third-party AI agents to access Citi Online or Citi Mobile.

The agreement says that if a customer gives an agent access, Citi will consider the agent’s actions authorised by the customer. It also says the customer is responsible for certain mistakes, unintended actions and actions that exceed the authority given to the agent, subject to applicable law.

Citi’s wording also addresses leaked credentials. If login details given to an agent are shared, leaked or distributed to someone else, the agreement says Citi will consider actions taken using those credentials authorised by the customer.

For example, if you gave an outside AI agent your Citi login and those details were later exposed, Citi’s agreement says it would treat activity using those details as authorised. Federal consumer protections still apply to qualifying unauthorised transfers, and the circumstances of the transaction matter.

Citi says this particular section does not apply to tools or features provided through Citi Online or Citi Mobile.

What Do Other Banks Say?

The agreements differ, so customers should read the terms for their own bank and account.

Chase

Chase’s digital services agreement says it can rely on instructions from an agent given access to its services as if the customer had provided them personally.

It also says customers are responsible for actions or errors resulting from an agent misunderstanding instructions, performing unintended actions or exceeding the authority given to it.

Wells Fargo

Wells Fargo’s upcoming online access agreement takes effect on 24 November 2026. It says customers are responsible for certain AI agent actions and errors, including unintended actions and actions beyond the authority given to an agent, unless otherwise provided by law or regulation.

The agreement also tells customers to revoke access promptly when they no longer want an agent to act for them.

PNC

PNC’s published digital services agreement, effective 19 December 2026, addresses third-party AI agents. It sets out customer responsibilities for certain activity and limits PNC’s responsibility for agent errors or unintended actions, subject to applicable law.

BMO

A BMO US deposit agreement update says instructions and transactions initiated by an AI agent can be treated as authorised by the customer. It also assigns customers responsibility for certain agent errors and losses, subject to applicable law.

The document reviewed by Digital Safety Squad did not clearly establish the effective date of the update. BMO customers should check the notices and terms provided for their account.

Bank of America

Bank of America has published guidance about connecting AI tools to financial accounts. Digital Safety Squad did not find a clause matching Citi’s explicit leaked-login wording in the online banking agreement reviewed for this article.

If you bank elsewhere, search your account agreement for ‘AI agent’, ‘third-party access’ and ‘account credentials’, or ask your bank how its rules apply.

Are Customers Responsible if Money Is Stolen?

Federal law protects consumers against certain unauthorised electronic transfers. A bank agreement cannot remove protections that apply under the law.

The US Consumer Financial Protection Bureau says an electronic transfer made using account information obtained through fraud or robbery can qualify as unauthorised. The rules also address situations where a customer gave another party access to an account. The facts, the type of transfer and the account involved are important.

This makes it important to distinguish between different situations:

If any of these happens, contact your bank immediately and explain exactly what happened. Ask the bank to investigate and keep a record of your report.

How to Protect Your Bank Account

1. Don’t give an AI agent your banking password

Your banking login can give someone access to sensitive information and, depending on your account, the ability to move money. Giving those details to an outside tool also means trusting another company to protect them.

Citi explicitly warns against using third-party agents to access its online banking services. Before connecting any tool, check your own bank’s rules and the permissions the service requests.

2. Check what the agent can access

An agent that can view information presents different risks from one that can make payments or change account settings.

Review the permissions before connecting a service. Give it access only to the information and actions needed for the task. If you can’t tell what it will be able to do, contact the provider before connecting it.

3. Keep control of payment approvals

Review the recipient, amount and account details before approving a payment. A tool can misunderstand an instruction, select the wrong recipient or act on incorrect information.

Where a service offers approval controls, turn them on and check whether they apply to every transaction.

4. Turn on account alerts

Transaction and login alerts can help you notice activity you don’t recognise sooner. The earlier you spot a problem, the sooner you can contact your bank and secure your account.

Review your statements as well. Reporting deadlines can affect your rights, and some depend on when the bank sends the statement showing the transfer.

5. Remove access you no longer need

If you stop using an AI agent, disconnect it using the options provided by the service and your bank. Check whether you also need to change your password or contact your bank to end access.

Keep a record of when you removed permission. Some bank agreements ask customers to retain evidence that they revoked an agent’s authority.

What to Do if an AI Agent Makes a Payment You Didn’t Want

Contact your bank immediately using the number on your card, its official app or the contact details on your statement.

Explain what happened, including whether you connected an AI agent, what you asked it to do and whether you approved the transaction. Ask the bank to investigate and tell you how to secure your account.

Save transaction details, messages from the agent and any record of the permissions you granted or removed. If your login details have been exposed, change them and follow your bank’s instructions.

If you’re unable to resolve the issue with your bank, you can submit a complaint to the Consumer Financial Protection Bureau.

Jade Artry

Jade Artry

Jade is a technology writer and editor with more than 16 years of experience helping people understand complex technology and what it means for their everyday lives. Her work has included years of hands-on testing and research across cybersecurity, online privacy, monitoring software, identity protection and digital safety, alongside more recent work examining the rapidly evolving world of consumer AI. Today, she focuses particularly on the safety, privacy and ethical questions surrounding AI, as well as its wider human impact: how these technologies are changing the way we think, communicate, trust, create and relate to one another.

View Articles