Datasets / Fraud Risk Management
Confirmed Fraud and Scam Labels
Investigator-confirmed fraud, scam and mule outcomes since 2022.
52 downloads
Updated weekly. Last update 2 days ago. Next expected in 5 days.
About this dataset
Investigator-confirmed outcomes for card fraud, account takeover, authorised push payment scams and mule accounts since 2022, linked to the originating transactions and accounts by token. Includes fraud typology, loss amount, recovery and the date confirmed. The ground truth for training and back-testing every fraud and mule model on the platform.
Usage rights
Group Financial Crime Compliance and approved model developers in Group Data & AI only. Customer-level data under banking secrecy (FSA 2013) and PDPA 2010. Stays in-region in Google Cloud Malaysia; export outside the region, re-identification attempts and joins to raw identifiers are prohibited. Direct identifiers are tokenised at ingestion; the token vault is held by Group Data & AI and is not exposed to consumers. Labels may not be used for any decision about an individual outside the fraud and AML case process.
Lineage
Exported weekly from the fraud case management system when a case is closed as confirmed or not fraud, joined to transaction and account tokens in silver. Label delay (median 11 days from transaction) is recorded per row so training windows can be set correctly. Dataplex checks label-to-transaction join coverage.
Ownership and sensitivity
Who approves access
- Owner, Fraud Risk Management, Chloe Tan
- Data Governance Office, Ahmad Faizal
Built on by Anti-Money Laundering.
Customer-level data under banking secrecy and PDPA, and models that decide on customers.
Named-user entitlement with the data owner's consent and a stated purpose; reviewed on expiry.
- Financial crime
- Used in AML, fraud or sanctions controls; tipping-off restrictions apply.
- Banking secrecy
- Customer account information protected under the Financial Services Act 2013.
- Customer PII
- Contains or processes personal data about customers (PDPA 2010).
3 models trained on this dataset
Details
- Updated
- 2026-08-01
- Cadence
- Weekly
- Format
- BigQuery table
- Licence
- BU Restricted
- Latest version
- 2.1