A Consensus-Driven News Verification System fighting misinformation through AI-powered multi-source analysis.

Links: Devpost · GitHub

Overview

FactNews is an advanced analytical platform that cross-verifies news information across multiple global media outlets. By employing a council of Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG), it identifies consensus, highlights contested facts, and provides transparent, evidence-based summaries of current events.

Key Features

• Multi-Source News Aggregation: Automated ingestion from 20+ major news outlets (BBC, Reuters, CNN, NYT, TechCrunch, etc.) via RSS feeds.
• AI Model Council: Multiple LLMs (OpenAI, Anthropic, DeepSeek, Google, etc.) deliberate in parallel, and a “judge” LLM synthesizes their conclusions to reduce individual model bias.
• Optimized RAG: Chunk-level semantic search with a multi-tiered embedding cache (Redis + local NPZ + API fallback) using fast NumPy vector operations.
• Bias & Divergence Analysis: Identifies where sources agree or disagree, highlighting potential media bias and showing contested facts with supporting evidence.

Tech Stack

• Frontend: Next.js 16, React 19, TypeScript, Tailwind CSS 4, Zustand
• Backend: FastAPI, Python, Uvicorn
• AI & Data: OpenAI, Anthropic, Google, DeepSeek, Redis, NumPy, RSS feeds