🛡 CyberSphere — The All-in-One Digital Trust Platform
HackThrone Project by Team CodeBlooded
🚀 Overview:
In today’s digital world, millions fall victim daily to phishing scams, fake news, and deepfake media.
Traditional tools only solve one part of the problem — some detect malware, others verify news — but there’s no unified platform that helps users verify online content easily and transparently.
CyberSphere bridges that gap.
It’s an AI-powered, explainable platform that allows users to:
Scan websites for phishing or scams 🕵♂
Verify headlines for fake or misleading content 📰
(Optionally) Detect deepfake images/videos 🎭
Learn about online safety and cyber hygiene 🔐
💡 Problem Statement:
Online deception — from scam websites and phishing links to fake news and deepfakes — has become rampant.
Users lack a single, easy-to-use verification tool that provides:
A trust score for any link, headline, or media
Human-readable explanations behind each verdict
Educational insights for better cyber awareness
This leads to financial frauds, misinformation spread, and reduced trust in digital spaces.
🌟 Opportunities:
🧩 All-in-One Verification: Website, News, and Media checks in one platform.
🧠 Explainable AI: Clear, human-readable reasoning for all predictions.
📈 Scalable Design: Expandable to browser extensions and mobile apps.
🎓 Educational Impact: Promotes digital literacy and cyber hygiene.
⚙ Methods & Approach:
Multi-Modal Detection Website Scanner:
Uses SSL validation, domain age, and blacklists.
Employs a lightweight ML phishing model.
Returns a Trust Score and reasoning.
News & Headline Checker:
NLP model using TF-IDF + Logistic Regression
Classifies as Real, Misleading, or Fake.
Deepfake Detector (Optional): Uses pre-trained AI models to analyze manipulated media.
Explainable AI Combines deterministic rules with AI predictions to generate transparent, human-understandable explanations for every result.
Architecture Frontend: React / Next.js
Backend: Flask / Node.js
Data Sources: PhishTank, Kaggle Fake News datasets, public threat intel
UI Tabs:
• Website Scanner
• News Checker
• Deepfake Detector
• Safety Insights
🔄 Process Flow:
User Input: URL, headline, or media file
Backend Processing: Sends input to relevant module (phishing, news, or media)
Detection Engine: Runs AI + heuristic analysis
Explainable Output:
Verdict (Safe / Risky / Fake / Real)
Reasons for decision
Educational insights and suggested next steps
🧩 Tech Stack:
Layer Technology
Frontend React / Next.js
Backend Flask / Node.js
AI Models TF-IDF, Logistic Regression, Heuristic Rules
Database SQLite / MongoDB
Datasets PhishTank, Kaggle Fake News Dataset
Optional Media Analysis Pre-trained Deepfake Detection Models
📊 Output Example:
Input Type Example Input Result Explanation
Website URL http://banklogin-secure.xyz ⚠ Suspicious Domain age < 30 days, SSL invalid, appears on blacklist
News Headline “NASA confirms aliens landed on Earth” ❌ Fake TF-IDF model similarity with known fake dataset
Image Manipulated celebrity video 🎭 Deepfake Detected Facial artifact mismatch and frame inconsistency
🧠 Learning Impact:
Educates users about scam patterns, fake news spotting, and cyber hygiene.
Encourages safe browsing habits and awareness of digital deception.
👥 Team CodeBlooded:
• Manan Singhal
• Viransh Jain
• Aagam Shah
• Smit Kamatnurkar
Our Final website can be viewed on: https://cyber-sphere-416372d1.base44.app
🏁 Future Scope:
🌐 Browser extension and mobile app versions
🗞 Real-time misinformation tracking
🧩 Integration with enterprise verification systems
🎓 Educational modules for schools and journalists
🏆 Hackathon: HackThrone
Built with ❤ by Team CodeBlooded to make the digital world safer, smarter, and more trustworthy.