# Citibank tells customers they are liable for losses if third-party AI tools access their accounts

> Citibank now holds customers liable for all losses caused by third-party AI tools they let access their accounts, even if credentials are stolen. The policy shift, matching moves by JPMorgan, Bank of America, and PNC, means the bank won't reverse unauthorized AI-driven transactions.

- Type: AI news
- Page: https://completeaitraining.com/news/citibank-tells-customers-they-are-liable-for-losses-if/
- Published: 2026-11-19
- Updated: 2026-11-19
- Author: Jeroen Erne
- Jobs: Customer support, Finance, Hospitality and events, Insurance, Legal, Sales
- Reading time: 3 min

Citibank updated its online user agreement this week to tell customers they are solely responsible for any losses, errors, or security breaches caused by third-party AI tools they authorize to access their accounts. The policy shift, which follows similar moves by JPMorgan Chase and Bank of America, puts the liability for automated financial assistants squarely on the consumer - even if the AI tool leaks their login credentials or gets hacked.

"As third-party AI tools become more widely available, we want customers to understand the risks associated with sharing banking credentials or authorizing those tools to access their accounts," a Citi spokesperson told TechCrunch. "We recently updated our Online User Agreement to provide greater clarity about those risks and to reinforce that customers who choose to grant account access to third-party AI agents are responsible for activities conducted through those agents."

#### Banks draw a hard line on AI access

Citi, the third-largest U.S. commercial bank, now warns customers that unauthorized transactions made by AI assistants - whether intentional or accidental - won't be reversed by the bank. The terms also cover credential leaks. If a third-party AI tool is breached and hackers use the stolen information to drain an account, the customer eats the loss.

JPMorgan and Bank of America, the country's two largest banks, announced comparable policies last month. PNC, the eighth-largest bank, changed its terms today. Citizens Bank sent a similar warning to its customers this week, according to an email reviewed by TechCrunch.

The message from financial institutions is consistent: connecting an AI agent to your bank account is a decision you make at your own risk. The banks will not serve as a safety net.

#### The tension between AI labs and service providers

AI startups see personal finance as a lucrative entry point. Agents that can analyze spending, suggest investment moves, or handle bill payments offer a direct line to transaction fees and consumer data. But the trust barrier is high. Consumers already hesitate to share email or chat records with AI tools. Handing over bank credentials raises the stakes considerably.

Anthropic's approach signals caution. The personal finance tool it launched earlier this month is "read-only" and, as the company put it, "can't move, manage, or invest your money." That limitation is deliberate - it sidesteps the liability question entirely by never touching funds.

Banks also view third-party AI agents as a competitive threat. Markets dipped briefly on the prospect of AI tools helping consumers dodge fees or shift money out of low-interest deposit accounts that prop up bank margins. Citi's updated terms make one exception: AI agents the bank itself provides come with no additional liability for customers, a clear signal that financial institutions intend to control the AI interface if it becomes a primary way people manage money.

This standoff mirrors a pattern playing out across industries. AI labs promise agents that will change how people interact with service providers. The service providers - from ride-hailing platforms to restaurant chains - push back with rule changes and access restrictions. How these conflicts resolve will determine which AI investments actually return value.

#### Why this matters for customer support, legal, and finance professionals

For customer support teams, expect a wave of calls from confused or angry customers who granted AI tools account access and now face unrecoverable losses. Support scripts will need clear, legally vetted language explaining the bank's position on third-party AI liability. Legal and compliance professionals should review their institution's terms of service against this emerging industry standard. If your bank hasn't updated its user agreement to address AI agents, that gap is a liability. For finance and insurance professionals, the policy shift creates a new risk category that clients may not understand - and one that standard fraud protections won't cover. Professionals who build or sell [AI Agent Courses](https://completeaitraining.com/tag/automation/) for financial use cases need to address these liability boundaries head-on, because customers will ask who pays when something goes wrong. The answer, increasingly, is the customer.

---

This is the Markdown form of https://completeaitraining.com/news/citibank-tells-customers-they-are-liable-for-losses-if/. The whole site is mapped at https://completeaitraining.com/llms.txt.
