OpenFraud Investigator

Trace the money. Connect the evidence. Investigate with AI.

OpenFraud Investigator is an open-source AI-powered financial fraud investigation assistant designed to help investigators understand complex transaction networks, explore suspicious financial activities and connect evidence across multiple accounts and beneficiaries.

Financial fraud investigations often involve examining large volumes of transaction records, identifying relationships between accounts, tracing potential money flows and gathering supporting evidence. OpenFraud Investigator addresses this challenge by combining graph analytics, deterministic fraud detection rules and an evidence-grounded AI copilot in a unified investigation workspace.

Key Capabilities

• Transaction Graph Analysis: Build interactive transaction networks to visualize account relationships, beneficiaries and potential money-flow paths.
• Fraud Pattern Detection: Identify selected suspicious patterns using configurable rules, transaction velocity analysis, multi-hop transfers and cross-currency comparisons based on recorded conversions.
• AI-Assisted Investigation: Use an AI copilot to explore cases, query transaction relationships and generate evidence-backed explanations.
• Evidence Management: Connect investigation findings to underlying transaction records and supporting evidence.
• Investigation Review: Support supervisor-reviewed investigation notes and maintain traceable investigation workflows.
• Responsible AI and Safety: Apply validation checks, access controls and safeguards to prevent unsupported AI conclusions and unauthorized information disclosure.

Technology Stack

• Frontend: Next.js, React, TypeScript
• Backend: Python, FastAPI
• Graph Analytics: NetworkX
• Database: PostgreSQL
• AI Integration: Open-weight LLMs through OpenRouter
• Visualization: Interactive transaction graph and investigation dashboard

Project Vision

To make complex financial investigations easier to explore, understand and document through open-source AI, graph intelligence and evidence-grounded workflows, while keeping human investigators in control of decisions.

Data and Safety: Uses synthetic banking data exclusively. This is a demonstration project, not a production banking system, and does not claim regulatory compliance or proven real-world fraud detection accuracy.