Newshot

Who's Liable When AI Agents Transfer Money or Shop on Their Own? Banks Warn Users May Bear the Loss

AI summary

Several major banks are warning customers that actions taken by AI agents with access to their financial accounts may be treated as the user's own actions, meaning users could be held responsible for losses. Ally Bank, JPMorgan Chase and Citi have all issued such warnings, which come as a new generation of AI agents gains the ability to shop and perform tasks autonomously, prompting retailers like Target to update their terms of service.

Translation(model translation, not the source's own text)

Who's responsible when AI agents transfer money or make purchases on their own? Multiple banks warn users may have to bear the losses themselves

IT之家 November 19 news: If your AI agent makes a bad financial decision, you may ultimately have to bear the loss yourself.

This is a warning that several major banks have recently issued to customers. According to bank emails and app notifications reviewed by Business Insider, banks are effectively reminding users that if an AI agent is granted access to financial accounts, the actions it performs may be treated as the user's own actions rather than operations carried out by an external intruder.

Ally Bank told customers in an email earlier this month: "When using AI services, you are responsible for any actions taken by the AI. AI may make mistakes, hallucinate, misunderstand your instructions, or execute instructions in ways you did not expect, causing significant losses."

JPMorgan Chase also reminded customers in a November email: "If your information is misused, authorized transactions may not be reversible."

Citi added a banner to its website and mobile app directing users to read the relevant risk disclosure. The page explicitly warns: "AI assistants can make mistakes, which means you may be responsible for the errors it causes."

These warnings are appearing one after another just as a new generation of AI agents is gradually gaining the ability to shop autonomously.

In the past, AI models such as OpenAI's ChatGPT and Anthropic's Claude mainly answered user questions through text. Today, online agents such as Meta's Muse, Instinct AI, and OpenAI's Dots can already perform practical tasks on users' behalf, such as booking flights, reserving restaurants, and even automatically placing orders before deodorant runs out.

Faced with these new tools, retailers are also beginning to adjust their terms of service. In March this year, Business Insider reported that Target revised its customer agreement to explicitly classify purchases made by AI agents on behalf of users as "transactions authorized by the user."

By September, 亚马逊, citing security and reliability concerns, banned Meta's Muse agent from making purchases on its website on users' behalf.

Banks are likewise trying to draw clear lines: connecting an account to an AI tool through a formal authorization method is not the same thing as handing login credentials directly to an AI tool.

Axios reported last month that several major banks had discovered third-party agents accessing customer accounts. Some of these operations were authorized by customers, and the AI agents also used login credentials provided by customers.

All the bank emails reviewed by Business Insider reminded customers not to provide important information such as usernames, passwords, and security verification codes to AI tools.

Wells Fargo also said customers should not readily hand account access to third parties.

A representative of the bank told Business Insider: "No technology should be allowed to take action on behalf of customers without strict security measures, clear customer authorization, and safeguards to protect customer funds and financial information."

In October this year, during JPMorgan's earnings call, the bank's CEO Jamie Dimon (Jamie Dimon) also issued a similar warning about AI agents accessing bank accounts.

He said: "If people are using these tools, they should understand the risks involved. In the meantime, we will not allow any external agent to freely enter the company's internal systems, because that would pose too great a risk to customers and to JPMorgan."

IT之家 noted that this campaign about AI security risks has also extended to device and software manufacturers.

In the beta version of Apple's latest-generation Mac operating system software, a related warning has already been added. When users consider granting an AI agent or application system-level access to the computer, the system will display a risk prompt reminding users to authorize cautiously.

Related articles

  • Apple warns: AI agent behavior may be unpredictable and could lead to lawsuits and damage the company's reputation
  • Cisco's president used an AI agent to inventory and cancel useless subscription services, saving thousands of dollars a year
  • To avoid a massive outflow of bank deposits, Citi and Bank of America step in to restrict agent access to customer accounts
  • Global financial regulatory standard-setting body calls on banks and companies to strengthen human oversight of AI agents
  • Production files for an entire season of a Swedish variety show were reportedly deleted by an AI agent, forcing the team to re-edit from scratch
  • Heavy investment in AI may not actually save money, McKinsey report says agents may expose companies to higher costs

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by IT之家
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗