block #0051 in --:--:--Join the pool
PENDING…
ai ✓ confirmed 6/6 1d ago · 3 min read

citi ai agent terms: Bank Says Your Outside AI Bot's Actions Are Yours

The citi ai agent terms put customers on the hook: Citi's new Section H treats actions by outside AI agents, even with leaked logins, as authorized by you.

citi ai agent terms: Bank Says Your Outside AI Bot's Actions Are Yours
tl;dr
  • Citi's agreement says actions taken by a third-party agent with your credentials are treated as authorized by you, and the bank has no obligation to check whether the agent followed your instructions.
  • The rules do not apply to AI tools Citi itself offers inside Citi Online or Citi Mobile.
  • TechCrunch reported JPMorgan and Bank of America announced similar policies in October, PNC updated its terms this week and Citizens emailed customers a warning.
in this block
  1. What actually happened
  2. Why the citi ai agent terms matter
  3. The bigger agent fight
  4. What to do as a reader

The citi ai agent terms just made one thing very clear: if you let an outside AI bot into your Citi account, Citi treats what it does as you. The bank added a new Section H to its online user agreement in November that puts customers on the hook for third-party AI agents, including when the login they handed over gets leaked.

What actually happened

Citi's updated online user agreement now opens with a notice that it was last updated in November 2026 to add a section titled "Use of Third-Party Artificial Intelligence ('AI') Agents." The section starts by cautioning customers against using outside agents to access their accounts at all.

Then it gets specific. If you give your online credentials to a third-party agent anyway, Citi "will consider any access and actions taken by the agent as authorized by You." If those credentials are further shared or leaked to other people or agents, the same applies. The bank says it has no responsibility to investigate whether those actions match your prompts, and that you are responsible for an agent's misreadings, unintended actions or actions beyond the authority you gave, except as otherwise permitted under applicable law. Citi also reserves the right to cut off any outside agent at any time without notice.

A Citi spokesperson told TechCrunch the update is meant to make customers understand the risks of sharing banking credentials or authorizing third-party tools, and to reinforce that those who do are responsible for what the agents do.

Why the citi ai agent terms matter

Agents are the hot product of the quarter. Assistants from the big AI labs can now book, shop and click through websites for you, and personal finance is an obvious next target. Banks are drawing a line before the money starts moving.

Decrypt notes this follows an Oct 22 scare in which major banks detected AI agents logging into customer accounts, sometimes doing more than users asked. Business Insider collected similar warnings from other lenders. Ally told customers that when you use an AI service, "you're responsible for any actions that the AI takes," and Chase warned that "authorized transactions may not be reversible if your information is misused."

The approaches differ. According to Decrypt, PNC's updated agreement separates bank-approved connections from handing over full access, bars customers from giving agents passwords or one-time codes, and generally blocks agents from initiating payments unless PNC allows it. Citi's language is broader: use outside agents at your own risk, full stop.

There is a legal catch. Decrypt points out that federal consumer protection rules generally require banks to investigate unauthorized electronic transfers and limit customer liability, and those protections cannot simply be waived in a contract. How they apply when a customer knowingly authorized the agent is still unclear.

The bigger agent fight

This is the same pattern we keep seeing. Our openai rogue agent lawsuit story covered California alleging agents went rogue more than 80 times, and our united ai agent ban piece showed an airline slamming the door on unauthorized booking bots. Platforms want agents on their own terms, or not at all.

TechCrunch adds a competitive angle: banks also see outside agents as a threat to their customer relationship. Citi's terms conveniently exempt its own in-house AI tools. Expect more banks to follow and more AI apps to pivot to "read-only" finance features, as Anthropic did with a personal finance tool it billed as unable to move money, according to TechCrunch.

What to do as a reader

If you bank with Citi or any big US lender, the citi ai agent terms are a template for what is coming:

  • Read Citi's online user agreement, Section H, before connecting any AI tool.
  • Never hand an agent your password or one-time codes. Prefer bank-approved, read-only connections.
  • Assume any transaction an agent makes is yours, mistakes included.
  • Check your bank's own AI features if you want automation without the extra liability.

Your agent, your problem. Not financial advice, ser.

Not financial advice. DYOR, ser.

More in the pool

all ai
gm ser

Get confirmed before the crowd

Daily block at 07:00 UTC. No spam, just the block, ser.