Logo Lanfrica

MrrAmissah/fraudcase-gh

Domaine:

peace and security

Type de record:

tools
Créateur:
Mrr
Hôte:
FraudCase Ghana is an AI-assisted fraud evidence organizer that helps individuals and small businesses collect, structure, analyze, and export scam-related evidence into a clean case report. # FraudCase GH > A privacy-first web app that helps people in Ghana organize digital-scam evidence, understand the risk, and produce a clean, non-accusatory incident report, without doxxing, shaming, or declaring anyone guilty. **Live:** fraudcase-gh.vercel.app FraudCase GH turns a confusing pile of suspicious SMS, WhatsApp messages, links, screenshots, and Mobile Money receipts into a structured case file with an AI-assisted risk assessment, **AI-extracted (and user-verified) evidence facts**: a timeline, an evidence checklist, and a downloadable PDF report suitable for sharing with a bank, mobile-money operator, or the National Cyber Security Authority. > **Positioning:** FraudCase GH is a decision-support and evidence-organization tool, **not** a law-enforcement system and **not** a verdict on any person. --- ## Table of contents - Problem statement - Key features - Tech stack - Architecture - Security & privacy model - AI safety & non-accusatory positioning - Local setup - Environment variables - Deployment - Roadmap - Further documentation --- ## Problem statement Digital fraud is widespread in Ghana: fake delivery/clearance-fee smishing (e.g. spoofed `GH-POST` senders), Mobile Money "wrong transfer / reversal" tricks, WhatsApp task-and-deposit recruitment scams, and credential-phishing links. When someone is targeted they face three problems: 1. **Disorganized evidence**: screenshots, texts, links, and receipts scattered and easy to lose. 2. **Uncertainty about risk**: hard to tell how dangerous a message is, or what to do next. 3. **Risk of overreaction**: public "name and shame" can defame innocent people and expose the victim's own data. FraudCase GH **organizes** evidence into a private case, **assesses** risk with cautious AI, and **frames** everything as non-accusatory decision support, while **redacting** sensitive data before any of it reaches a model. --- ## Key features - **Quick Check (public, no sign-up).** Paste a suspici …