Wells Fargo customers could be liable for their own AI agents' actions, according to interpretations

U.S. bank Wells Fargo has updated its online banking agreement, the Online Access Agreement. The new terms take effect on November 24, 2026, and cover, among other things, AI tools and AI agents. An AI agent is a program that can act on its own, for example by logging in and performing tasks for a user.

What does the agreement say?

According to the bank's own page, a third-party tool or AI agent that a customer uses to access their accounts or services is not operated by Wells Fargo.

If customers give others access to their accounts, they are responsible for actions taken through that access. This also applies to actions taken through Wells Fargo's own AI tools. The bank also reserves the right to restrict or block access from certain third-party AI tools.

How far does the liability extend?

A legal commentary concludes that customers who authorize an agent can be held responsible for the transactions the agent initiates. According to the commentary, this also applies to unintended activity and actions that go beyond the customer's instructions, subject to applicable law. Wells Fargo reserves the right to restrict or block access from certain third-party AI tools.

The German tech outlet BornCity writes that customers will be liable going forward for all actions, errors and transfers made by AI agents if they have given the agent their login credentials. According to the outlet, the bank can require prior registration, limit access rights and keep read access separate from payment functions.

The legal commentary also points out that contract terms remain subject to federal law. That includes the Electronic Fund Transfer Act and Regulation E, which limit consumer liability for certain unauthorized electronic transfers. According to the commentary, the consumer protection agency CFPB has not yet explicitly addressed how the rules apply when AI agents exceed the customer's instructions. It is therefore unclear how far the customer's liability extends in practice.

Other banks are tightening up too

According to BornCity, several large U.S. banks are introducing restrictions. Morgan Stanley blocks external agents entirely, while Citi prohibits agent logins and transactions but allows approved read access through open banking providers.

What does it mean for the reader?

The case primarily concerns Wells Fargo customers, and other banks have their own restrictions, according to BornCity. But it highlights a question that more people may face: if you let an AI agent use your online banking at Wells Fargo, the bank may take the view that you are the one who bears responsibility if something goes wrong.