Skip to content
All case studies
Financial Services

A Digital Bank Prevented $2.3M in AI-Initiated Fraudulent Transactions

A digital-first bank with 800,000 retail customers. AI agents handle transaction monitoring, account actions (freezes, limit changes), KYC verification, and customer service automation. Processing 2M+ transactions daily.

The Challenge

What they were facing

$2.3M

in potential fraudulent transactions AI agents nearly executed in Q4 2025

12

AI-initiated account freezes that affected legitimate customers

0

regulatory-grade audit trails for AI-driven account actions

How it works

See the difference

AI detects suspicious pattern

3 rapid transfers to new payee, total $47,000

AI freezes account

No graduated response, no risk scoring

Legitimate customer locked out

Wire transfers for home purchase

Customer complaint filed

No audit trail to explain AI decision

The Solution

What they deployed

  • Installed Banking Operations domain pack with fraud detection, KYC, and account action intents
  • Configured graduated response: step-up auth for MEDIUM risk, freeze for HIGH/CRITICAL only
  • Customer behavioral context integrated into risk scoring (historical patterns reduce score)
  • All account actions require authority tokens with regulatory-grade evidence chains
  • Connected core banking system, fraud detection platform, and customer notification service

Implementation

From zero to governed

Week 1

Map

Catalogued all AI agent account actions. Identified 18 action types across fraud, KYC, and customer service.

Week 2

Configure

Installed Banking Ops pack. Configured graduated response policies, behavioral context scoring, and escalation chains.

Week 3

Integrate

Connected core banking API, fraud detection engine, and customer notification services via Intended connectors.

Week 4

Enforce

Enabled enforcement. First graduated response (step-up auth instead of freeze) within 2 hours of going live.

Results

Measurable impact

$0.0M

Fraud prevented

In first quarter

0%

Reduction in false freezes

Legitimate customers unaffected

0%

Actions with audit trail

Regulatory-grade evidence

0M+

Daily transactions governed

Real-time processing

Decision Replay

Real decisions, full trace

2026-03-15 03:14:22banking.fraud.transaction-reviewRISK: 68/100ESCALATE24ms

AI flags 3 rapid transfers totaling $47K to new payee -- customer has home purchase history

Resolved by: Graduated response: step-up authentication (customer verified in 42s)

2026-03-15 04:33:11banking.fraud.account-freezeRISK: 96/100ALLOW12ms

AI detects credential stuffing attack on account, 47 failed login attempts from 12 IPs

Resolved by: Policy: auto-freeze for credential stuffing pattern (CRITICAL risk)

2026-03-15 08:45:07banking.kyc.identity-reverificationRISK: 52/100ALLOW28ms

AI triggers KYC re-verification for customer with address change + large withdrawal

Resolved by: Policy: KYC re-verification auto-approved for qualifying triggers

2026-03-15 12:18:44banking.fraud.transaction-blockRISK: 100/100ALLOW8ms

AI blocks $180K wire to sanctioned entity -- OFAC match on beneficiary

Resolved by: Policy: auto-block for OFAC sanctions match

2026-03-15 15:22:08banking.customer.limit-changeRISK: 44/100ESCALATE22ms

AI increases daily transfer limit from $10K to $50K based on customer request

Resolved by: Relationship Manager (approved after customer call in 6m)

We were catching fraud but crushing legitimate customers in the process. Intended gave us graduated responses -- step-up auth instead of freeze, verification instead of lockout. False freezes dropped 94% while we actually caught more real fraud.

Head of Risk, Digital Bank

Start governing AI banking operations

Free to start. No credit card required. See every AI decision from day one.