Your Real-Time Guardian Against Misinformation
AI-Powered Browser Extension for Truth & Transparency Across the Web
TrustMeter combines 5 AI models, and semantic caching to verify text, detect fake images, and score website credibility — instantly.
Every verification request flows through multiple AI systems for maximum accuracy
Text/Image/URL
Pinecone + Firestore
Fact check API + Custom search + Vertex AI Model + Gemini and More
Trust Rating
We believe truth should be accessible to everyone. TrustMeter brings enterprise-grade fact-checking, image verification, and credibility scoring right into your browser — making safe and informed browsing effortless.
Instantly detect misleading claims using AI ensemble: Gemini AI, Vertex AI ML, and Google Fact Check.
Detect fake or AI-generated visuals using Google Cloud Vision and custom Vertex AI models.
Get instant credibility scores for any website using Firestore domain scoring and semantic evidence matching.
Accurate verification powered by async parallel processing, Pinecone + Firestore caching, and concurrent API calls across all verification systems.
Combines 6 systems: Gemini 2.5 Flash (primary reasoning), Vertex AI (ML classification), Google Vision API (image analysis), Fact Check API, Sentence Transformers (semantic matching), and Google Custom Search.
Real-world corroboration > ML predictions. System prioritizes professional fact-checks and multi-source evidence.
User feedback weighted by unique user counts and stored in Pinecone with 15-day auto-expiry. Firestore tracks domain reputation with rolling averages based on verification outcomes, hence making the system more accurate over time.
SHA-256 hash match → Semantic similarity → 'Known fake' database. Pinecone vectors expire after 15 days; scoring weighted by unique user confirmations of similar news.
Firestore-powered credibility database that updates in real-time. Domains earn reputation through verification results. High-trust sources (90%+ score, 100+ votes) receive confidence bonus.
5-phase async architecture: Fact-check + metadata extraction → Vertex AI + corroboration → Multi-claim analysis → Ensemble scoring → Dual storage (Firestore + Pinecone).
Weighted scoring system: Professional fact-checks override ML predictions. Evidence count scales confidence. Context-aware temporal validation.
Separate image verification pipeline using Google Vision API for reverse image search, metadata extraction, Vertex AI image analysis with Gemini Fallback hence providing visual content analysis.
Intelligent pre-filtering distinguishes factual claims from opinions, personal stories, satire, questions and promotional content. Returns "Not Applicable" for non-verifiable content types.
Reformulating claims into alternate phrasings for comprehensive search coverage. Combined with keyword extraction for maximum evidence discovery.
System evaluates claims against current date context. Distinguishes between timeless facts and time-sensitive events.
Extension monitors the webpage for text, images, and links that may need verification.
The Cache is checked for similar content that was recently verified.
If not cached, run parallel analysis through Gemini, Vertex AI, Vision API, Fact Check API, Custom search corroboration and much more.
Search trusted sources using Google Custom Search with dynamic keyword extraction via Gemini.
Weighted combination: Gemini + Vertex AI + Fact Check + Custom search produces final verdict.
Results displayed with transparent reasoning. Stored in Pinecone and Firestore for future queries.
Misinformation affects daily decisions, from what we buy to what we believe. By putting AI-powered truth verification directly into your browser, TrustMeter builds a safer digital world where information is Trustworthy, Verifiable, and Transparent.
The misinformation crisis demands intelligent, real-time solutions
In today's digital landscape, misinformation spreads faster than ever before. Deepfakes, manipulated images, and biased narratives are increasingly difficult to detect.
With billions of posts shared daily, manually fact-checking every claim is impossible for the average user.
AI-generated deepfakes and edited media are becoming indistinguishable from authentic content.
False information spreads 6 times faster than the truth on social media platforms.
Algorithms amplify misleading content, creating echo chambers and reinforcing biases.
Users often lack context about source credibility and content authenticity.
Traditional fact-checking is time-consuming and doesn't scale for real-time browsing.
TrustMeter brings enterprise-grade verification tools directly to your browser, empowering you to make informed decisions in real-time without the need for manual fact-checking.
Our system works like having a team of expert fact-checkers analyzing every piece of content you see—except it happens instantly. By combining multiple AI models with professional fact-checking databases and real-world evidence from credible sources, TrustMeter delivers accurate verdicts in seconds, not hours. The platform doesn't just tell you if something is misleading; it explains why with clear reasoning, provides links to supporting evidence, and shows you the credibility of the sources involved. As users interact with the system and verification results accumulate, TrustMeter continuously learns and improves, building a comprehensive reputation database for news sources and content patterns. Whether you're scrolling through social media, reading news articles, or encountering viral images, TrustMeter acts as your intelligent companion—protecting you from misinformation while helping you understand the truth behind the headlines.
Built on robust, scalable technology for reliable verification
TrustMeter combines cutting-edge AI technology with enterprise-grade infrastructure for reliable, real-time verification
Chrome Extension - Manifest V3 with content scripts, popup UI, background service workers, and real-time DOM analysis
Flask API - Async Python backend with ThreadPoolExecutor, parallel processing pipelines, and RESTful endpoints
Gemini 2.5 Flash (reasoning & structured analysis)
Vertex AI (custom text classification)
Google Vision API (image features verification)
Sentence Transformers (semantic embeddings)
Pinecone - Vector database for semantic caching, similarity search, and 15-day auto-expiry
Firestore - NoSQL document store for domain reputation, user feedback, and metadata
Google Fact Check Tools API - Professional fact-check aggregation
Google Custom Search API - Evidence corroboration and source discovery
sentence-transformers - all-MiniLM-L6-v2 for text embeddings
google-generativeai - Gemini SDK
aiohttp - Async HTTP client
python-dotenv - Environment configuration
LRU Caching - In-memory embedding cache (8192 items)
Async/Await - Concurrent API calls with asyncio.gather()
Extension monitors webpage for text, images, and links. User triggers verification via context menu or button click.
Pinecone semantic search and firestore queries for similar content. If found with similarity(cosine and verified fakes search) and user confirmation, return cached result.
If cache miss, trigger parallel processing: Gemini extracts metadata, Vertex AI classifies, Vision analyzes images.
Google Fact Check API searches for professional fact-checker ratings and information. Gemini refines claim for better matching.
Google Custom Search with dynamic keywords and summaries finds supporting/contradicting evidence from trusted domains.
Firestore domain scores updated based on verification results. Weighted average by unique user votes.
Final result stored in Pinecone (vector) and Firestore (document) with 15-day TTL.
Findings sent back to extension and displayed in user-friendly overlay with explanation.
Install TrustMeter in minutes and start browsing with confidence
Click the button below to download the latest version of TrustMeter from GitHub.
Unzip the downloaded file. You'll find an /app folder containing all extension files.
Navigate to chrome://extensions/ in your Chrome browser's address bar.
Toggle the Developer mode switch in the top-right corner of the extensions page.
Click Load unpacked and select the /app folder you extracted earlier.
The TrustMeter icon will appear in your browser toolbar. Click it to activate verification on any webpage.
Highlight any text on a webpage and right-click, then select "Verify with TrustMeter" to check for accuracy and bias using 5 AI models.
Right-click on any image and choose "Check Image Authenticity" to detect manipulation or AI generation using Google Vision + Vertex AI.
Click the TrustMeter icon in your toolbar to get an instant credibility score for the current website using Firestore domain scoring.
Ensure Developer mode is enabled in chrome://extensions/ and you've selected the correct /app folder.
Check your internet connection and ensure the backend API is running. Try refreshing the webpage and check browser console for errors.
Clear your browser cache and disable other extensions temporarily. First-time checks are slower; subsequent checks use Pinecone cache.
Verify all API keys are correctly set in your .envFIREBASE_KEY_PATH, GOOGLE_API_KEY, GOOGLE_SEARCH_CX, GEMINI_API_KEY, PINECONE_API, FIRESTORE_CREDENTIALS_PATH, GOOGLE_APPLICATION_CREDENTIALS, PROJECT_ID, IMG_ENDPOINT_ID, TEXT_ENDPOINT_ID, LOCATION, GOOGLE_VISION_AI_CLIENT_SECRET, GOOGLE_VISION_AI_CLIENT_ID, APP_SECRET_KEY
Check Pinecone index status and ensure Firestore rules allow read/write. Run cleanup_expired() if cache is stale.
Provide feedback through the extension to improve accuracy. System learns from unique user votes and updates domain scores.
Open source on GitHub. Contributions, issues, and stars welcome!
github.com/akshaynair5/H2S_hackathon
Built for the GenAI Hackathon by passionate developers committed to fighting misinformation with cutting-edge AI.
Contributors: Akshay Nair, Gokul Krishna and Sarath Rajendran.
Need help? Have suggestions? Found a bug?
Get your API keys and configure TrustMeter in minutes
Only if you wanted to run server locally
TrustMeter uses several AI services to provide accurate verification. You'll need to create accounts and obtain API keys from Google Cloud Platform and Pinecone. Most services offer free tiers that are sufficient for personal use.
Service: Google Gemini AI
Cost: Free tier available (60 requests/minute)
Steps:
1. Visit Google AI Studio
2. Sign in with your Google account
3. Click "Get API Key" → "Create API Key"
4. Copy the generated key
Service: Pinecone Vector Database
Cost: Free tier (100K vectors)
Steps:
1. Visit Pinecone.io
2. Sign up for a free account
3. Go to "API Keys" in the console
4. Create a new index (dimension: 384, metric: cosine)
5. Copy your API key and index name
Services: Vision API, Custom Search
Cost: $300 free credit for new users
Steps:
1. Visit Google Cloud Console
2. Create a new project
3. Enable these APIs: Vision API, Custom Search API
4. Go to "Credentials" → "Create Credentials" → "API Key"
5. Copy the API key
Service: Google Vertex AI ML
Cost: Included in Google Cloud free tier
Steps:
1. In Google Cloud Console, enable Vertex AI API
2. Go to "IAM & Admin" → "Service Accounts"
3. Create service account with "Vertex AI User" role
4. Create JSON key and download it
5. Note your Project ID
Service: Google Fact Check Tools API
Cost: Free (no quota limits)
Steps:
1. In Google Cloud Console
2. Enable "Fact Check Tools API"
3. Use the same API key from step 3
4. No separate key needed
Service: Google Custom Search API
Cost: 100 queries/day free
Steps:
1. Visit Programmable Search Engine
2. Create new search engine
3. Set to search entire web
4. Copy your Search Engine ID
5. Use Google Cloud API key from step 3
Service: Google Cloud Firestore
Cost: Free tier (1GB storage)
Steps:
1. Visit Firebase Console
2. Add Firebase to your Google Cloud project
3. Enable Firestore Database
4. Use the same service account JSON from step 4
5. Note your Project ID
Fill in your credentials below — TrustMeter will create a complete .env file template for you.
Double-check that you've enabled the correct APIs in Google Cloud Console and that your keys have no extra spaces.
You may have hit the free tier limit. Check your usage in the respective consoles or upgrade to a paid plan.
Ensure your service account has the correct roles: Vertex AI User, Firestore User, and Storage Object Viewer.
Make sure service-account.json is in your project root directory, not in a subdirectory.