AI Bodyguard

AI Bodyguard: Real-time scam detection & safety alerts. Your digital shield for risky calls, messages, and meetups. Stay protected

Comment

Project name: AI Bodyguard Author: Edcaly Laguerre Hackathon: 305 Hackathon — Smart City Innovation (Fall 2025)

Description AI Bodyguard is a community-driven safety platform that detects scams and protects users across dating, job hunting, and peer-to-peer marketplaces — scanning profiles, chats, calls, SMS, and live meetups, and storing crowdsourced reports to continuously improve detection. Problem Statement Online interactions (dating apps, job postings, marketplace listings) are rife with scams, harassment, catfishing and real-world safety risks. People lack a simple, real-time guard that (1) flags risky profiles/messages/calls, (2) helps ensure safe meetups, and (3) crowdsources scam intel to protect the broader community. Solution Overview AI Bodyguard is a full-stack demo that: • Scans user profiles and assigns a risk score. • Protects conversations by flagging scam/harassment signals. • Simulates call and SMS scam detection. • Enables Safe Meetups with check-ins and emergency escalation (simulated). • Provides a public scam reporting portal to collect data and train models. • Stores all data in MongoDB Atlas; supports optional Google Gemini for richer AI analysis or runs in mock mode for hackathon demos. Key Features (as delivered)

  1. Profile Scan (POST /api/user/profile-scan) o Input: username, email, (optional) profile pic url o Output: riskScore (Safe or High) — mocked/randomized in demo mode; stored in Users collection.
  2. Chat Protection (simulated via frontend) o Real-time flagging workflow simulated in demo UI.
  3. Safe Meetup o Start meetup: POST /api/meetup/start → stores meetup o Check-in: POST /api/meetup/checkin → updates status (safe / alert) o On alert, backend logs: Emergency! Notify authorities for user X (simulation)
  4. Scam Call Detector (POST /api/scam/call) o Submit call transcript/phone → mocked Safe/Scam response; stored in ScamLogs.
  5. Scam SMS Detector (POST /api/scam/sms) o Submit SMS content → mocked Safe/Scam response; stored in ScamLogs.
  6. Public Scam Reporting Page (POST /api/scam/report) o Fields: scamType, reporterName, email, phone, description → stored in ScamReports.
  7. Scammer Intelligence DB o Store phone/email/profile images and logs to help future detection.
  8. Optional AI Analysis Endpoint (if API key present) o POST /api/scam-analyze → forwards report text to Google Gemini (or other model) and returns analysis. If no key, mock analysis runs. Tech Stack • Frontend: React (CRA) — components: ProfileScan, ScamReport, SafeMeetup, ScamCall, ScamSMS • Backend: Node.js + Express • Database: MongoDB Atlas (cloud) — collections: Users, ScamReports, Meetups, ScamLogs • Optional AI: Google Gemini (via server-side API), or any LLM via environment key. • Hosting / Dev: Meku / Replit / Glitch / Vercel (pick one for live demo) • Integrations: Twilio (optional for real SMS/call workflows), Mapbox (optional for safe location suggestions) • System Architecture (summary) • User (browser/mobile) → React frontend → Express API → MongoDB Atlas Optional: Express → Google Gemini API for advanced NLP analysis Emergency flow: meetup.checkin with alert -> backend simulates contacting authorities (console log / webhook placeholder).
Share this project:

Updates