Describe the AI system
Sign in to save your workDescribe the system in plain language before judging it. Use only what the case tells you and mark anything else as unknown.
BankFraud AI at Northbridge Bank
- Provider
- Sentinel Analytics Ltd (fictional supplier) builds and maintains the scoring model
- Deployer
- Northbridge Bank configures the thresholds and operates customer review
- Inputs
- Amount, merchant category, location, device, time, customer spending history
- Output
- Fraud-risk score 0–1 with an action: approve, verify or block
- Thresholds
- 0.60–0.85 verify by app or SMS; above 0.85 block and send an alert
- Human role
- Fraud analysts review blocks only when a customer calls or the urgent queue escalates
- Live since
- Fourteen weeks; the supplier updates the model quarterly
AI system description
This becomes the first section of your report.
Name the system
What problem is the organisation trying to solve, and how will it know the system helps?
Provider, deployer, users, affected people, suppliers, system owner, risk owner and control owners
Prediction, classification, recommendation, generation, retrieval or action; classic or generative AI
List what the evidence confirms, what you are assuming and what you do not know
Which decision does the output influence, and what does a person still decide?
Where, for whom and for what the system may and may not be used
Answers save automatically and sync to your account when you are signed in.