Citi Puts AI-Agent Hack Losses on Customers; 5 Banks, $9.6T

Citibank changed its terms of service this week to tell customers who use third-party AI tools for personal finance that anything that goes wrong is their problem, even if the tool is hacked, TechCrunch reported on 18 November. PNC changed its terms the same day, and Citizens emailed customers a similar warning.

Our number: the five banks TechCrunch names with such policies, JPMorgan, Bank of America, Citi, PNC and Citizens, held about $9.6 trillion of consolidated assets at their US bank units on 30 June, per the Federal Reserve’s large commercial banks list (our sum).

Business Pill · Least Privilege

An evergreen Business Engineer explainer on least privilege and access control. It teaches the general idea only and does not discuss this report.

The key insight: Citi now tells customers who give third-party AI agents account access that they are responsible for what the agents do, even if the tool is hacked, TechCrunch reports; with JPMorgan, Bank of America, PNC and Citizens, the banks named hold about $9.6 trillion of assets by our sum of Fed data.

What Citi Changed

A Citi spokesperson told TechCrunch: “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”.

The spokesperson added that Citi “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”.

According to TechCrunch, Citi’s update also tells customers that if an AI tool leaks their credentials or is hacked, the customer is responsible for anything hackers do with the account. TechCrunch adds that Citi’s terms note that AI agents Citi itself provides can be used without that additional liability.

Consolidated assets of the US bank units of the five banks TechCrunch names with AI-agent liability policies,
Consolidated assets of the US bank units of the five banks TechCrunch names with AI-agent liability policies, as of 30 June 2026: JPMorgan Chase Bank $4.09 trillion, Bank of America $2.65 trillion, Citibank $1.98 trillion, PNC $610 billion and Citizens $233 billion. The about $9.6 trillion total is our sum. Source: Federal Reserve Large Commercial Banks release.

Which Banks Have Moved

TechCrunch reports that JPMorgan and Bank of America announced similar policies last month, PNC changed its terms on 18 November, and Citizens emailed customers this week warning that they are on the hook for anything their AI assistant does, according to an email TechCrunch saw.

By the Fed’s 30 June data, the five bank units range from JPMorgan Chase Bank at $4.09 trillion to Citizens Bank at $233 billion; Citibank alone held $1.98 trillion. The first wave of US bank moves in late October and the UK banks that followed were mostly about blocking agent access; this round is about who pays when an agent the customer authorised goes wrong.

Why It Matters for Agentic Finance

TechCrunch notes that Anthropic’s personal finance tool is billed as “read-only” and “can’t move, manage, or invest your money”. It also reports that banks see agents as a competitive threat as well as a security one.

The same week, BNY and Microsoft ran an AI agent buying Treasury bills, Singapore set up a sandbox for agents that pay out on contract terms and the Fed’s Michael Barr warned agents could speed bank runs. Consumer agents get a narrower path: the bank’s own agent, or the customer’s own risk.

The Structural Read

As we read TechCrunch’s report, the policy shift is from blocking agents to pricing them: customers may connect a third-party agent, but the loss sits with them.

As we read Citi’s carve-out for its own agents, the liability rule also steers customers toward the bank’s own AI, which TechCrunch frames as banks fighting to control that pathway.

A Citi spokesperson, to TechCrunch

“customers who choose to grant account access to third-party AI agents are responsible for activities conducted through those agents”

Three Implications

FOR CUSTOMERS TechCrunch reports Citi makes them responsible even when an AI tool leaks credentials or is hacked.

FOR AI FINANCE APPS TechCrunch notes Anthropic’s personal finance tool is billed as read-only and cannot move money.

FOR BANKS TechCrunch reports Citi’s own AI agents can be used without the additional liability.

The Business Engineer Lens

This story maps onto the Business Engineer framework The Agentic Harness War.

The analysis puts it this way: “The app stops being a destination and becomes a capability the harness invokes.”

As we read the banks’ new terms, they are refusing that role for third-party agents: an outside agent can invoke the account, but the bank will not stand behind what it does. The analysis predates this report and does not discuss banks.

What Is Not Established

We could not read Citi’s or PNC’s updated agreement text directly; the terms are described here as TechCrunch reports them and as Citi’s spokesperson put them. We did not see the Citizens email. The $9.6 trillion is the assets of the banks’ US bank units, not the number of customers who use AI agents, which none of the sources gives.

This Week In The AI Supercycle

AI Lost Its Central Banker, the Lab Oligopoly, and the Future of Open Source

This week’s issue: the Nvidia leadership change, read as the AI build-out losing its central banker: who now finances the GPU landlords, backstops the data centres and sponsors open models, and why two labs are becoming the bank.

Read the full issue →

Business Engineer Framework

The Agentic Harness War

A Business Engineer analysis of where value pools as AI agents act across the apps people already use.

Read the Map of AI →

The Bottom Line

Citi joined JPMorgan, Bank of America, PNC and Citizens in putting AI-agent losses on customers, including losses from a hacked tool, at banks holding about $9.6 trillion of assets, by our sum of Fed data.

97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.

A note on sourcing. Bank policies, the Citi spokesperson’s statement and the Anthropic tool description are from TechCrunch (18 November 2026), read in full; we could not read the agreements themselves. Assets are consolidated assets of each bank’s US bank unit from the Federal Reserve’s Large Commercial Banks release (data as of 30 June 2026). The $9.6 trillion total ($9,564 billion) is our sum. We did not contact the banks.

Sources: TechCrunch, ‘More banks says customers using AI agents are on their own’, 18 Nov 2026 · Federal Reserve, Large Commercial Banks, data as of 30 Jun 2026

Scroll to Top
FourWeekMBA