Siasat

Dont share, verify first

Comment

Overview

Siasat is a web-based fact-checking tool that helps users verify the content they consume online. It supports YouTube and TikTok video analysis, AI-generated image detection, and a Chrome extension for scanning news articles and text-heavy sites like Reddit. By extracting transcripts, scanning images, and evaluating written content, Siasat returns credibility assessments so users can make informed decisions about what they see and read.

Problem

Misinformation spreads rapidly online — a false claim can go viral within minutes. Users who lack the time or background to independently verify what they see are especially vulnerable. Current fact-checking solutions are often too slow to keep pace with content creation, and many require research skills that casual viewers don't have.

Solution

Siasat addresses these gaps by offering three core capabilities:

  • Real-time video analysis — Extracts and evaluates claims from video transcripts as they're submitted.
  • Confidence scoring — Assigns a confidence level to each claim so users can gauge reliability at a glance.
  • Supporting context — Provides explanations and source references alongside every verdict, giving users the information they need to judge for themselves.

Features

  1. YouTube Video Scanner:
    • Extracts the video's transcript via URL, identifies individual factual claims, and returns a verdict with confidence scores for each one.
  2. Video Scanner
    • Works the same way as the YouTube scanner — upload a video to pull the transcript and get a claim-by-claim fact-check.
  3. Image Scanner:
    • Upload an image to detect whether it was generated by AI. Returns a classification with a confidence score.
  4. Article Scanner (Chrome Extension)
    • A browser extension that scans the current page for misinformation risk. Works on news sites, blogs, Reddit, and any other text-based content on the web.

Built With

Share this project:

Updates