Everything you need to know about how FraudVerse AI works.
It screens bank transactions and accounts for fraud and mule-account activity using a trained ML model, explains every prediction with SHAP feature attributions, and maps out connected accounts using graph intelligence to surface fraud rings.
A mule account is a bank account used to receive and move money obtained through fraud — often opened by someone recruited (knowingly or not) to pass funds along, making the money harder to trace back to the original crime.
A trained classifier scores each account/transaction using behavioral and transactional features (velocity, device changes, transfer patterns, etc). SHAP values then break that score down into the specific factors that pushed it up or down.
Some workspaces (like Investigation case details) use illustrative data because the live backend doesn't expose that endpoint yet. Dashboard, Predictions, Graph, and Reports pull from the real pipeline.
See our Privacy Policy for the full breakdown of what's processed, stored, and never shared.
FraudVerse AI was built by team Vyuh for a national hackathon, focused on explainable, graph-aware fraud detection for public sector banks.