AI CyberShield – Real‑Time Scam & Fraud Detection System

• 

  AI CyberShield landing page with live demo preview and feature overview.

• 

  Analyzer dashboard for pasting suspicious messages and running instant scans

• 

  Scan results dashboard showing scam probability, risk score, and live threat feed status.

• 

  Explainable AI highlights with confidence band and pattern matches for transparency

• 

  Detailed URL reputation analysis and model category scores

• 

  Threat history dashboard with weekly blocked scams, cumulative risk saved, and reports.

Inspiration

Scams and phishing are no longer limited to emails — they hit users through SMS, social media, crypto wallets, and everyday browsing. We wanted to build something that protects people in real time, not just after they paste a message. That vision led to AI CyberShield: a proactive security system rather than a reactive tool.

What it does

AI CyberShield analyzes suspicious messages, URLs, and browsing activity to detect scams. It outputs a scam probability, risk score, and explainable reasons, and the browser extension provides real‑time alerts, link highlighting, and optional auto‑block protection.

How we built it

• Frontend: Next.js + Tailwind dashboard
  
• Backend: FastAPI with AI scoring + rule detection + threat intel
  
• Extension: Chrome Manifest V3 for live scanning
  
• AI: Hosted NLP model (Hugging Face) + rules engine
  
• Threat feeds: URLhaus + VirusTotal
  

Risk is computed by combining multiple signals:

$$ \text{Risk} = 0.4 \cdot \text{AI} + 0.2 \cdot \text{Patterns} + 0.4 \cdot \text{URL/Feeds} $$

Challenges we ran into

• Free threat feeds often rate‑limit or block requests
  
• Making the extension fast without spamming the API
  
• Balancing strong detection with privacy‑first storage
  
• Turning model signals into clear, user‑friendly explanations
  

Accomplishments that we're proud of

• Real‑time browser protection with offline fallback
  
• Explainable AI highlights and confidence bands
  
• Automated threat scoring that feels like a real product
  
• Privacy‑first storage and consent flow
  

What we learned

Security isn’t just about detection accuracy — it’s about trust, explainability, and UX. We learned how to combine AI, threat intelligence, and browser‑level protection into one cohesive experience.

What’s next for AI CyberShield – Real‑Time Scam & Fraud Detection System

• Add more threat feeds with reliable caching
  
• Train a custom model focused on crypto & financial scams
  
• Expand to mobile/PWA scanning
  
• Build a community reporting network that strengthens detection over time
  

Built With

• chrome-extension-manifest-v3
• fastapi
• hugging-face-inference-api
• javascript
• next.js
• node.js
• npm
• python
• sqlite
• tailwindcss
• typescript
• urlhaus
• uvicorn
• virustotal