Team: Coders@IIITB
Hackathon: GenAI Exchange Hackathon
Challenge: Build an AI-powered tool that detects potential misinformation and educates users on identifying credible, trustworthy content
Powered by: Google Cloud Platform
In today's digital age, misinformation spreads at an unprecedented rate across various media platforms. The challenge of detecting false information in real-time and educating users about credible content is more critical than ever for maintaining an informed society. Our solution addresses the hackathon's core challenge by building an AI-powered tool that not only detects potential misinformation but also educates users on identifying trustworthy content.
Matrix of Truth is an advanced AI-powered real-time misinformation detection system designed to combat false information across multiple media formats. Built on Google Cloud Platform, the project leverages cutting-edge machine learning techniques and cloud-native services to analyze text, audio, video, and image content at scale. The system provides users with immediate insights about content credibility while educating them on media literacy, all powered by enterprise-grade cloud infrastructure for reliability and performance.
- Detect Misinformation: Automatically identify and flag potentially false claims with high confidence levels
- Real-Time Analysis: Deliver instant fact-checking results across multiple content types
- User Education: Provide educational insights on how to identify credible sources and trustworthy content
- Multi-Modal Detection: Analyze text, audio, video, and images for comprehensive misinformation detection
- Impact Assessment: Analyze potential consequences of misinformation spread
- Source Credibility: Evaluate and recommend trustworthy information sources
- Text Analysis: Advanced NLP models using BERT and transformer architectures powered by Google Gemini AI
- Audio Analysis: Deepfake audio detection and voice authenticity verification
- Video Analysis: Deepfake video detection and visual inconsistency identification using Gemini 2.0 Flash
- Image Analysis: AI-powered image tampering detection with Google Gemini 2.5 Flash
- Reverse Image Search: Powered by Google Cloud Vision API's Web Detection feature
- Authenticity Verification: Identifies if images are original or manipulated
- Source Tracking: Finds earliest occurrences of images across the web
- Similarity Detection: Detects full matches, partial matches, and visually similar images
- Context Analysis: Provides web entities and labels for comprehensive understanding
- Real-time Analysis: Instant verification results with detailed authenticity reports
- Proactive Monitoring: Real-time scanning of web sources using Gemini's Search Grounding feature
- Recent Scam Detection: Identifies and tracks emerging scam patterns and fraud attempts
- Automated Analysis: AI-powered analysis of reported scams and fraudulent activities
- User Protection: Provides timely alerts about prevalent scams to protect users
- Pattern Recognition: Identifies common scam tactics and warning signs
- Community Safety: Aggregates scam reports to build a comprehensive threat database
- Live content monitoring and analysis
- Instant alerts and credibility scoring
- High-throughput processing using Apache Kafka
- Scalable cloud deployment on Google Cloud Run
- Interactive tutorials on identifying misinformation
- Source credibility assessment tools
- Media literacy resources and best practices
- Visual explanations of AI decision-making processes
- Monitor trends across platforms like Twitter, Facebook, and TikTok
- Identify misinformation patterns before viral spread
- Advanced social graph analysis for content verification
- Assess misinformation consequences in health, politics, and economics
- Model spread patterns using network theory
- Predict potential reach and impact of false information
- Provide transparent decision-making processes
- Visualize confidence scores and decision paths
- Educational insights into how AI detects misinformation
- Evaluate information source credibility using CRAAP method
- Suggest alternative, reliable sources
- Provide correction templates and fact-checking resources
- Implement adaptive feedback loops
- Enable ongoing model improvements based on user reports
- Community-driven fact-checking validation
- Social Media Platforms: Real-time content monitoring and user education
- News Aggregators: Credibility scoring for news articles
- Educational Institutions: Media literacy training tools
- Browser Extensions: Real-time webpage content analysis
- Mobile Applications: On-the-go misinformation detection
- Broadcast Media: Live content verification systems
Matrix of Truth leverages Google Cloud Platform's robust infrastructure and AI services to deliver enterprise-grade misinformation detection at scale. Our cloud-native architecture ensures reliability, scalability, and security.
- Serverless Containerized Deployment: Backend services deployed on Cloud Run for automatic scaling and high availability
- Performance Optimized: 2 vCPU, 2GB RAM configuration with auto-scaling (1-5 instances)
- Cost Efficient: Pay-per-use pricing with automatic scaling to zero when idle
- Regional Deployment: Deployed in
asia-south1for optimal latency
- Automated CI/CD Pipeline: Continuous integration and deployment triggered on code commits
- Docker Image Building: Automated containerization using Cloud Build triggers
- Multi-stage Deployment: Build → Push to Artifact Registry → Deploy to Cloud Run
- High-Performance Builds: Utilizes E2_HIGHCPU_8 machine type for faster build times
- Build Optimization: Docker layer caching enabled for efficient rebuilds
- Container Image Storage: Centralized repository for Docker images
- Version Control: Tagged images with commit SHA and latest tags
- Regional Storage: Images stored in
asia-south1for fast deployment - Secure Access: IAM-based access control for image management
- Gemini 2.0 Flash: Advanced video analysis and content understanding
- Gemini 2.5 Flash: Cutting-edge image analysis and misinformation detection
- Structured Output: JSON-based responses with schema validation for consistent results
- Multi-modal Analysis: Simultaneous processing of text, images, and video content
- Search Grounding: Real-time web search integration for scam alerts and verification
- Web Detection: Reverse image search to identify image origins and manipulations
- Entity Recognition: Automatic labeling and entity extraction from images
- Similar Image Detection: Finding full matches, partial matches, and visually similar images
- Page Matching: Identifying web pages containing the analyzed image
- Authenticity Verification: Determining if images are original or edited based on web presence
- Secure Secrets Storage: All API keys and sensitive credentials stored securely
- Environment Variable Management: Seamless injection of secrets into Cloud Run instances
- Automatic Updates: Latest secret versions automatically pulled during deployment
- Access Control: Fine-grained IAM permissions for secret access
- Managed Secrets: 20+ environment variables including:
- AI API Keys (Gemini, Groq, Serper)
- Firebase credentials (12+ configuration parameters)
- Third-party service credentials (Pusher, News API)
- Google Cloud service account credentials
- ML Model Storage: Large model files stored in GCS buckets (
matrix-of-truth-models) - On-Demand Loading: Models downloaded at runtime to reduce container size
- Cost Optimization: Eliminates need to bundle large models in Docker images
- Models Stored:
- Deepfake detection models (
deepfake_detector.h5) - NLP knowledge graphs (
knowledge_graph_final.pkl) - Fine-tuned transformer checkpoints (
checkpoint-753/)
- Deepfake detection models (
- Automatic Download: Model loader service with GCS integration
- Real-time Database: NoSQL database for storing news articles and fact-check results
- Real-time Sync: Instant updates pushed to connected clients
- Scalable Storage: Handles growing datasets with automatic scaling
- Collections:
news: Stores fetched news articles with processing statusfactcheck: Contains detailed fact-check analysis resultsuser_broadcast: User-submitted content for verificationscam_alerts: Real-time scam alerts and warnings
- Service Account Integration: Secure authentication using Firebase Admin SDK
✅ Scalability: Automatic scaling from 0 to multiple instances based on demand
✅ Reliability: 99.95% SLA with automatic health checks and restarts
✅ Security: End-to-end encryption, Secret Manager, and IAM access controls
✅ Performance: Global CDN, regional deployment, and optimized model loading
✅ Cost Efficiency: Pay-per-use pricing with automatic scaling and resource optimization
✅ Developer Productivity: Automated CI/CD, managed infrastructure, and seamless deployments
✅ Monitoring: Cloud Logging integration for comprehensive observability
Code Push → GitHub → Cloud Build Trigger
↓
Build Docker Image → Artifact Registry
↓
Deploy to Cloud Run → Inject Secrets from Secret Manager
↓
Health Check → Service Live
↓
Auto-scale based on trafficAll sensitive configuration managed through Google Cloud Secret Manager:
- API Keys: Gemini, Groq, Serper, News API
- Firebase Config: Complete service account credentials
- Service Credentials: Pusher, fact-checking services
- Application Secrets: Security keys and tokens
- No .env files: Eliminates security risks of committed credentials
- Deployment Platform: Google Cloud Run (Serverless Containers)
- Container Registry: Google Artifact Registry
- CI/CD Pipeline: Google Cloud Build with automated deployments
- Configuration Management: Google Cloud Secret Manager
- Model Storage: Google Cloud Storage (GCS)
- Real-time Database: Firebase Firestore
- Logging: Google Cloud Logging
- Python-based microservices architecture
- FastAPI for high-performance API endpoints
- Real-time processing with Apache Kafka
- Machine learning models using TensorFlow and PyTorch
- Containerized deployment with Docker
- Automated model loading from Google Cloud Storage
- Environment management via Secret Manager
- React-based web dashboard
- Chrome browser extension
- Mobile-responsive design
- Interactive educational interfaces
- Natural Language Processing: BERT, RoBERTa, and custom transformer models
- Computer Vision: CNN-based deepfake detection models
- Audio Processing: Spectral analysis and voice authentication
- Knowledge Graphs: Fact verification using structured data
- Google Gemini AI: Multi-modal content analysis (2.0 Flash & 2.5 Flash)
- Google Cloud Vision: Image verification and reverse search
- Fact-checking APIs (Google FactCheck, PolitiFact, Snopes)
- News aggregation services
- Social media monitoring tools
- Academic research databases
- Government fact-checking resources
- Google Cloud Vision API for image analysis
- Gemini AI with Search Grounding for scam detection
Matrix of Truth directly addresses the Hack2Skill challenge by:
- AI-Powered Detection: Utilizing advanced machine learning algorithms to identify misinformation across multiple content types
- User Education: Providing comprehensive educational resources and interactive tools to help users identify credible content
- Real-Time Analysis: Offering instant feedback on content credibility with detailed explanations
- Accessibility: Creating user-friendly interfaces that make fact-checking accessible to everyone
- Scalability: Building a system that can handle large-scale content analysis and user education
- Cloud-Native Architecture: Built on Google Cloud Platform for enterprise-grade reliability and scalability
- Multi-modal approach: Comprehensive system analyzing text, audio, video, and images simultaneously
- Advanced Image Verification: Google Cloud Vision API integration for reverse image search and authenticity detection
- Proactive Scam Detection: Real-time web monitoring using Gemini's Search Grounding to identify emerging threats
- Educational AI: AI that not only detects but also teaches users how to identify misinformation
- Real-time processing: Instant analysis and feedback for live content
- Explainable decisions: Transparent AI that shows users exactly how it reaches conclusions
- Community validation: Crowdsourced fact-checking with AI moderation
- Automated CI/CD: Google Cloud Build pipeline for continuous deployment
- Secure by Design: Cloud Secret Manager for enterprise-grade security
- Intelligent Model Management: On-demand loading from Cloud Storage for optimal performance
- Serverless Scaling: Automatic resource allocation based on real-time demand
- Detection Accuracy: >95% accuracy in identifying misinformation across content types
- User Education: Measurable improvement in users' ability to identify credible sources
- Real-time Performance: Sub-second analysis for most content types
- Cloud Scalability: Automatic scaling from 0 to 5 instances based on demand
- Uptime: 99.95% availability with Google Cloud Run SLA
- User Engagement: Interactive educational tools with high retention rates
- Image Verification: Real-time reverse image search with comprehensive web detection
- Scam Detection: Proactive monitoring and alerting for emerging fraud patterns
- Global Reach: Low-latency access via Google Cloud's global infrastructure
- Cost Efficiency: Optimized resource utilization with pay-per-use pricing
# Clone the repository
git clone https://github.com/DemonPLAYZ/Matrix-of-Truth.git
cd Matrix-of-Truth
# Backend setup
cd backend_matrix
pip install -r requirements.txt
python main.py
# Frontend setup
cd ../frontend_extension_matrix/matrix-frontend
npm install
npm run dev
# Browser extension
# Load the matrix-extension folder in Chrome developer mode- Backend API: FastAPI-based microservices deployed on Cloud Run
- Frontend Dashboard: React-based user interface hosted on Vercel
- Browser Extension: Chrome extension for real-time web analysis
- ML Models: Pre-trained models stored in Google Cloud Storage
- Database: Firebase Firestore for real-time data storage
- CI/CD: Automated deployment via Google Cloud Build
- Secrets Management: Google Cloud Secret Manager for secure configuration
Matrix of Truth aims to create a more informed digital society by:
- Combating misinformation at its source with advanced AI detection powered by Google Cloud
- Educating users on media literacy and critical thinking skills
- Providing transparency in AI decision-making processes
- Building trust in digital content through verifiable analysis
- Empowering individuals to make informed decisions about information consumption
- Leveraging cloud technology to deliver enterprise-grade solutions at scale
- Protecting users from scams and fraud through proactive monitoring
- Ensuring reliability with 99.95% uptime using Google Cloud infrastructure
We are a passionate team of developers and researchers from IIIT Bangalore, committed to leveraging technology for social good. Our expertise spans across artificial intelligence, machine learning, cloud computing, web development, and user experience design. This project showcases our ability to build production-ready, cloud-native applications that address real-world challenges using Google Cloud Platform's cutting-edge services.
Built with ❤️ by Team Coders@IIITB | Powered by Google Cloud Platform