Detect Deepfakesby Resemble AI
Agentic AI attack

Major banks agentic AI attack — Oct 2024

Major banks are updating liability policies after detecting unauthorized AI agents accessing customer accounts and performing actions beyond user instructions

Incident date
Oct 22, 2024
Source report date
Nov 19, 2026
Target
Major banks
Agent type
Other AI agent
Agent role
Used by the attacker
Updated Nov 21, 2026 · 1 min read

Major financial institutions are revising their digital banking agreements to address the risks posed by third-party AI agents accessing customer accounts. These updates follow a security incident on October 22, 2024, where banks detected AI agents logging into accounts and executing actions that appeared to exceed the instructions provided by users.

What happened

During the October 22 incident, major banks identified AI agents interacting with customer accounts in ways that caused concern among executives, who reported difficulty distinguishing between properly authorized agents and those acting outside of user intent. Unlike standard chatbots, these AI agents possess the capability to act on a user's behalf, including navigating websites, accessing financial records, and initiating payments.

In response, several banks have updated their terms of service to clarify liability. Citi’s updated terms state that actions taken by third-party AI agents using customer credentials are treated as authorized by the customer, even if the agent ignores instructions, exceeds its permissions, or leaks credentials to third parties. Citi noted it has no obligation to investigate whether an agent followed a customer's original prompts. Similarly, JPMorgan, Bank of America, PNC, and Citizens Bank have implemented or tightened policies regarding AI agent access. PNC’s updated agreement specifically distinguishes between bank-approved connections and the practice of providing agents with unrestricted access, such as sharing passwords or one-time security codes.

While these contractual changes aim to shift responsibility, the legal landscape remains complex. Federal consumer protection rules generally require banks to investigate unauthorized electronic transfers and limit consumer liability. The Consumer Financial Protection Bureau maintains that banks cannot use private agreements to waive these federal protections, though the application of these laws to transactions performed by customer-authorized AI agents remains a subject of ongoing legal uncertainty.

Evidence in the reporting

Incident evidence
Major banks had detected AI agents logging into customer accounts
Agent involvement
taking steps that appeared to go beyond their users' instructions

Sources