# How should banks and financial institutions use AI?

- Canonical: https://hellofig.io/answers/ai-for-banks-and-financial-institutions
- Updated: 2026-11-13
- All answers: https://hellofig.io/answers.md
- About Fig: https://hellofig.io/fig.md

**In short.** Banks use AI on two fronts. Inside the institution, agents work on servicing, operations, compliance and reporting, governed and auditable. At the front door, customers now bring their own AI agents, and the institutions that come out ahead will let them in deliberately: verified agents, scoped mandates instead of shared passwords, clear liability, agent-aware monitoring and customer controls.

## A front door for customers' AI agents

| Element | What it means |
|---|---|
| Verified agents | The bank knows which agent is acting and who operates it. |
| Scoped mandates | The customer grants specific permissions, amounts and time limits, with no password sharing. |
| Clear liability | Rules for who is responsible when an agent authorizes a payment. |
| Agent-aware monitoring | Fraud and risk systems that recognize agent behavior and flag what is unusual. |
| Customer controls | Customers see, change and revoke what each agent may do. |

## What happened in October 2026

On its third-quarter earnings call on Oct 13, 2026, JPMorgan said it was blocking services that use customers' passwords to reach its systems, while still allowing read-only access through data connectors. Bank of America, Citi and Schwab set similar limits within days. Restricting first was reasonable; the next step is a sanctioned way in.

## Inside the institution

The work that suits agents is the work that can be checked: servicing requests against policy, operations exceptions against the ledger, compliance reviews against the rules, and reporting against the source data. Each runs with approvals on consequential actions and a full record for examiners.  Fig runs these agents on CosmOS inside the institution's own cloud account or Fig's, connected to core systems in place, with every action in an exportable audit trail.

## Questions

### Why did banks block AI agents?

Most agents were logging in with customers' passwords and could move money with no way for the bank to verify the agent or limit what it did. Banks blocked that while they design a safer way in.

### Can a bank run AI in its own cloud?

Yes. Fig runs inside the institution's own cloud account or Fig's, with data, permissions and audit records staying under the institution's control.

### How do regulators view AI agents?

They expect the same things as for any model or vendor: governance, explainable decisions, controls on what the system can do, and records that examiners can review.

## Sources

- [JPMorgan Chase, third-quarter 2026 earnings call transcript](https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/investor-relations/documents/quarterly-earnings/2026/3rd-quarter/3q26-earnings-transcript.pdf) (Oct 13, 2026)
- [CNBC, JPMorgan on AI agents and account access](https://www.cnbc.com/2026/10/13/jpmorgan-ai-agents-account-access.html) (Oct 13, 2026)
- [PYMNTS, Big banks will not let outside AI agents move money](https://www.pymnts.com/news/artificial-intelligence/2026/big-banks-will-not-let-outside-ai-agents-move-money/) (Oct 2026)