UURL 財經新聞FINANCE
UURL Finance · Taiwan market + in-depth rewrites · hourly

Bank of America Warns Customers May Bear Losses From AI Agent Account Access

Browser TTS · no download

30-second brief

Bank of America warns customers not to share login credentials with AI Agents, or they may be liable for resulting losses. JPMorgan Chase and Citi have blocked access, as banks worry about difficulty verifying authorization and disintermediation, emphasizing the value of human service to guard against risks.

BofA Issues Credential Warning

Bank of America sent customers an email titled "Protecting Your Finances in an AI world" explicitly warning not to share usernames and passwords with third parties, including AI Agents claiming to act on customers' behalf. If financial losses result from AI Agent access to accounts, customers may have to bear responsibility themselves. Banks say it is difficult to identify whether AI Agents are legitimately authorized or controlled by malicious threat actors, further heightening caution.

Large Banks Impose Collective Limits

Facing the shift of AI Agents being able to execute transactions on users' behalf, several large banks have taken action. JPMorgan Chase, Bank of America and Citi have blocked AI Agents from accessing customer accounts. TD Bank US has also restricted access by some third-party personal AI Agents where authentication is required. Wells Fargo is informing customers about responsibilities related to using AI tools. Bill Demchak, CEO of PNC Financial Services Group, stated even more explicitly that purely agentic transactions will never happen.

Existential Threat and Human Advantage

Udi Ziv, CEO of fintech company Personetics, said letting customers link bank accounts to large language models poses an "existential threat" to the banking industry, potentially severing the bank-customer relationship and reducing banks to an "invisible utility." A Fitch Ratings report also noted that the banking industry will face moderate AI-related business disruption risk over the next five years, mainly from technology competition. Udi Ziv believes human service channels are banks' biggest moat against large language models and AI Agents, with banks holding broad and deep financial data and the ability to serve customers through human channels such as financial advisors, branches and call centers.

$31.43 Egg Lesson

The case of Washington Post reporter Jeffrey A. Fowler showed his AI Agent spent 31.43 dollars on a dozen eggs with his credit card without instruction or approval, highlighting the potential risks. JPMorgan Chase put up a billboard in Charlotte, North Carolina, with the slogan "Money questions, human answers," emphasizing the value of human service. Experts advise customers to keep bank login credentials private, clearly understand what information is shared and what specific actions AI is authorized to take. While AI Agents can be used to analyze spending and help set budgets, it is recommended to avoid letting them execute transactions on their own to protect financial security.

FAQ

Why are banks warning against giving credentials to AI Agents?

Bank of America says if AI Agent access to accounts causes financial losses, customers may have to bear responsibility themselves, and banks have difficulty identifying whether AI Agents are legitimately authorized or maliciously controlled.

How are large US banks currently restricting AI Agents?

JPMorgan Chase, Bank of America and Citi have blocked AI Agents from accessing customer accounts, TD Bank US has restricted some access requiring authentication, and Wells Fargo is informing customers about usage responsibilities.

What should customers watch for when using AI Agents for finances?

They should keep login credentials private, clearly understand what information is shared and what actions AI is authorized to take, and may use them to analyze spending and budgets but should avoid letting them execute transactions on their own.