Citi Makes Customers Liable for Third-Party AI Agents, Even if Their Login Credentials Leak

Citi has added a new Section H to its Online User Agreement. If customers give a third-party AI agent their login, Citi treats everything the agent does as authorized by the customer, including actions after leaked credentials, and says it has no duty to investigate.

Christopher Ort
By i10X News Desk · Reviewed by Christopher Ort · 3 min read
Share
A person at a kitchen table at night looks at a banking app on a smartphone next to a laptop.
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. Background
  5. Why it matters
  6. What's still open
  7. About the analysis
  8. i10x Perspective

Citi has added a new Section H, “Use of Third-party Artificial Intelligence (‘AI’) Agents,” to its Online User Agreement for Citi Online and Citi Mobile. The bank cautions customers against letting outside AI agents into their accounts and states that if they do, Citi will treat everything the agent does as authorized by the customer, even when the agent misreads instructions or the login credentials leak further.

i10X Quick Take

Quick Take

  1. What changed: Citi’s Online User Agreement was updated in November 2026 with a new Section H on third-party AI agents.
  2. Authorization: Any access and actions by an AI agent that has a customer’s credentials are treated as authorized by the customer.
  3. Leaks: If those credentials are “further shared, leaked, or otherwise distributed,” actions taken with them also count as authorized.
  4. No duty to check: Citi says it has no obligation to investigate such actions, even if they conflict with the customer’s prompts.
  5. Exceptions: The terms do not apply to AI tools Citi itself provides, and customer responsibility applies “except as otherwise permitted under applicable law.”

In plain English

If you hand your Citi password to an AI assistant so it can pay bills or move money for you, Citi now says in writing that whatever the assistant does is on you. That includes mistakes, actions you did not intend and what happens if the assistant leaks your password and someone else uses it. Citi’s own AI features inside its apps are not covered by this rule.

What happened

Section H defines an AI agent as “software or other technology that takes autonomous, semi-autonomous, or non-autonomous action on behalf of, or at the instruction of, You.” It states: “You are cautioned against using any third-party AI Agent to access your accounts or use any digital services.”

Customers who ignore that caution are responsible for “any actions or errors that result from the third-party AI Agent’s incorrect interpretation or execution of your instructions, unintended actions performed by the third-party AI Agent, or actions taken by the third-party AI Agent that exceed the authority granted.” The agreement also makes monitoring the agent the customer’s job, reserves Citi’s right to block any agent “at any time without notice,” and notes that the Citi Privacy Notice does not apply to the third-party agent.

Background

Banks have been tightening rules for AI agents that log in with customer credentials for weeks, as we reported when large banks began restricting agent access. Regulators are also looking at the issue: Singapore’s central bank is planning a know your agent protocol so banks can verify who an agent represents.

Why it matters

Consumer AI agents are being marketed as tools that can manage money, but banks are pushing the risk squarely onto users. Citi’s wording goes further than a warning by covering leaked credentials and disclaiming any duty to investigate conflicting actions. Until banks offer sanctioned, verifiable agent access, using a general-purpose AI agent with a bank login means accepting liability for its mistakes. How these terms interact with US consumer protections for unauthorized transfers remains to be tested.

What’s still open

  • How courts and regulators will treat agent-initiated transactions under existing consumer protection rules.
  • Whether Citi will offer an approved, read-only or limited-permission path for third-party agents.
  • Whether other banks adopt similar language on leaked credentials.

About the analysis

This independent i10x analysis is based on the text of Citi’s updated Online User Agreement. Quotes are taken from that agreement.

Sources (plain text, no links):

  • Citi Online User Agreement, Section H “Use of Third-party Artificial Intelligence (‘AI’) Agents” (updated November 2026)

i10x Perspective

The agentic economy needs an identity and liability layer before it can touch money at scale. See which institutions deploy AI in our Banks Using AI module and on the Finance page.

Christopher Ort

Christopher Ort

Christopher Ort is the editor of i10X News at i10x.ai. He has worked hands-on with AI tools, models and agent workflows for years and leads i10X’s coverage of AI launches, funding and research. He is the author of “The Style Penalty”, an SSRN study on how frontier AI models judge AI-written CVs, covered by heise online and t3n.

Editor's picks

  1. Dragos canary study: attackers stole AI prompts and credentials at three AI inference providers
  2. Analysis: AI Agents in Banking Hit a Liability Wall
  3. CVE-2026-133631: FlowIntel chatbot let LLM call any API, exposed Ollama token in process table
  4. Stripe Evals: Coding Agents Score 78% on Payment Integrations, but Only 29% Pass Every Required Check
  5. Kong Volcano: A Self-Healing Agent Cloud Where Coding Agents Build, Host and Fix Apps