Policy & Regulation

Citi shifts legal risk of AI agents onto its own customers

Citi's November 2026 user agreement makes customers liable for all third-party AI agent activity, even when leaked credentials put unknown actors inside the account.

Citi Makes Customers Responsible for Third-Party AI Agent Activity
Citi Makes Customers Responsible for Third-Party AI Agent Activityviarami / pixabay
By Sophie Lindqvist4 min read

Why it matters

  • Citi's November 2026 Citi Online User Agreement update, reported by TechCrunch on November 18, 2026, makes customers responsible for third-party AI agent activity.
  • Access using agent credentials that are shared, leaked, or distributed is still treated as customer-authorized.
  • Citi can limit, suspend, or terminate any third-party agent's access at any time without notice.
  • The Citi Privacy Notice does not apply to third-party agents receiving sensitive customer data.
  • Section 4 of the funds-transfer agreement deems agent-initiated transfers directly authorized by the customer.

By Sophie Lindqvist

Citi's November 2026 update to its Citi Online User Agreement makes customers legally responsible for everything third-party AI agents do inside their accounts — including actions taken with credentials that later leak to someone else entirely. TechCrunch reported the revised terms on November 18, 2026.

The change arrives as banks and fintechs brace for a wave of agentic AI tools that log into accounts, move money, and act on a user's behalf. Citi's answer is to draw a hard line: the bank won't referee what an agent did or why. The customer who let the agent in owns the consequences.

What does the new Section H actually say?

The new Section H kicks in whenever third-party AI agents or "agentic AI" access accounts or digital services through Citi Online or Citi Mobile. Citi defines an AI agent broadly — software or other technology that takes autonomous, semi-autonomous, or non-autonomous action for a customer, or for someone acting on the customer's behalf. One group is carved out: agents, tools, functions, and features Citi itself provides through those channels.

That definition matters because it sweeps in far more than fully autonomous bots. A半-autonomous assistant that suggests a payment and executes it on confirmation counts. So does software a family member or advisor deploys on your behalf.

Citi explicitly cautions customers against using third-party agents to access accounts, provide information or instructions, or conduct transactions. But if a customer goes ahead anyway — supplying online credentials or otherwise permitting access — the bank will treat the agent's access and actions as authorized. The warning, in effect, is the liability.

The takeaway: under Section H, granting an agent access means the bank treats whatever follows as something you asked for.

What happens if agent credentials leak?

Here the terms get unusually aggressive. If credentials provided to an agent are subsequently shared, leaked, or otherwise distributed to other people or other AI agents, access and actions using those credentials are also treated as customer-authorized.

Citi states it has no responsibility or obligation to investigate the authority or validity of those actions — even when they conflict with the customer's actual prompts or other instructions. An agent that misreads a request, does something unintended, or exceeds the authority the customer granted is the customer's problem, except as otherwise permitted under applicable law.

Monitoring sits on the customer's side of the ledger too. The agreement makes checking an agent's activity for accuracy, completeness, and compliance with the granted authority the customer's responsibility, not the bank's.

In plain terms: if your agent's login leaks and a stranger drains the account, the contract says that looks like you authorizing the transfer.

Who controls the agent's access — and your data?

The responsibility flows one way, but the control does too. Citi reserves the right to limit, prevent, suspend, or terminate access by any third-party AI agent at any time, without notice. The bank can also restrict which routes an agent takes into its digital services. Even so, contractual responsibility for activity already permitted stays with the customer.

The data provisions follow the same pattern. Granting an agent account access authorizes Citi to disclose customer information — including sensitive personal information — to that agent. The Citi Privacy Notice does not apply to the third-party agent, and Citi says it does not control how that agent collects, uses, stores, protects, or shares customer data.

So one decision — letting an agent in — simultaneously opens the door to sensitive data and closes the privacy notice that would otherwise govern it.

Do money transfers follow the same rules?

Yes, through a separate document. Citibank's Agreement for Online Funds Transfers carries its own November 2026 update notice addressing third-party AI agents. Its Section 4 makes customers solely responsible for instructions supplied to Citi by an agent acting on their behalf.

Agent-initiated transfers are deemed initiated and directly authorized by the customer, without further investigation by Citi. That holds even when the instructions exceed or conflict with the prompts or other actions the customer actually took with the agent.

The two agreements draw a consistent distinction: a customer's prompt is one thing, but the authority created by granting account access is another — and it's the authority that binds.

Why this matters beyond Citi

For personal-finance AI providers, Citi has just set a demanding accountability model. Their products can be cut off without notice, while their users carry responsibility for monitoring agent conduct and for transfers the bank treats as directly authorized.

The structure also creates a gap banks may find convenient: the institution can block agents at will, yet disclaim any duty to verify that an agent's actions matched what the customer actually asked for. Whether that allocation survives scrutiny under consumer-protection law is an open question — but until it's tested, Citi customers delegating to AI agents are, contractually, flying solo.

Source: unite.ai

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.

777 articles

Related articles

  1. Four banks make customers liable for third-party AI agent losses
  2. Incode's Know Your Agent Ties Every AI Agent to a Verified Human
  3. Uptiq launches GovernAI to govern AI agents in banking at no extra cost
  4. Bank of America warns customers not to share passwords with AI agents
  5. JPMorgan, Amazon and Kohl's Are Blocking AI Agents. Experts Say They Can't Win That Fight

« Previous articleNext article »