Author: Nishtha Mehrotra | EQMint

Synopsis:

• Major US banks, including JPMorgan Chase, Bank of America, Morgan Stanley and Wells Fargo, are responding to security risks involving third-party AI agents accessing customer accounts.
• Reports indicate that some autonomous agents have attempted actions beyond user instructions, raising concerns about unauthorised transactions and fund transfers.
• Banks are blocking or restricting access as they struggle to distinguish legitimate customer activity from actions performed by autonomous AI systems, even when users have shared their credentials.
  
  

Major US banks are tightening restrictions on third-party artificial intelligence agents accessing customer accounts, citing concerns about unauthorised activity and financial security. An October 22 report by Axios highlighted the challenges banks face as customers increasingly use AI tools to automate financial tasks.

Morgan Stanley has reportedly begun blocking third-party AI agents, while JPMorgan Chase, Bank of America and Citigroup are also restricting access. Wells Fargo was among the institutions identified in reports concerning these security risks.

The issue arises when customers share their banking credentials with AI assistants to automate financial tasks. Although users may authorise access, banks face difficulties distinguishing legitimate customer activity from actions performed autonomously by AI systems.

Why AI Agents Pose a Security Challenge

AI agents can perform tasks on behalf of users, but their actions may sometimes extend beyond the instructions provided. This raises concerns about unintended purchases, fund transfers and unauthorised account activity.

Banks also face challenges in verifying whether an action was initiated directly by a customer or executed by an autonomous system. Even when customers voluntarily share their credentials, institutions may be unable to reliably distinguish between the two.

However, the reports do not establish that customer funds were stolen or that the incidents involved confirmed account takeovers.

Read-Only Access Could Offer a Middle Ground

One possible solution is read-only access, which would allow AI tools to retrieve financial information without authorising transactions. Customers could potentially use these tools for budgeting, spending analysis and financial planning while retaining control over their money.

Similar access models are already used by some financial applications, including Quicken. Developing secure authentication and permission systems could help banks distinguish between viewing account information and executing transactions.

The developments highlight a growing challenge for the financial sector as autonomous AI tools become more capable. Banks must balance demand for automation with customer security, regulatory obligations and clear accountability when an AI agent acts unexpectedly.

Disclaimer: This article is for information purposes only and is not investment advice.

For more such information, visit EQMint

Join our WhatsApp channel for timely updates: Whatsapp