Skip to content

denis-furtuna/ScamSlap-bot

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🛡️ ScamSlap - The Anti-Deepfake Shield for Messenger

A digital bodyguard designed to protect your family from AI-generated frauds (Deepfakes) directly on Facebook Messenger.


📸 Multilingual Force Demonstration

ScamSlap is built with intelligence: it automatically detects the language set on the user's phone and responds in their native tongue. No configuration needed—the weapon calibrates itself!

Here is ScamSlap in action, intercepting and analyzing media in 3 different languages.

🇷🇴 Romanian 🇺🇸 English 🇪🇸 Spanish

(Real Messenger screenshots showing the analysis and interception process)


🔥 Why ScamSlap? (Our Mission)

Deepfake scams (fake videos of celebrities, cloned voices, AI-generated images) are becoming increasingly difficult to detect with the naked eye, especially for the elderly. ScamSlap provides an instant line of defense.

Did Grandpa receive a suspicious Reel? He just "Shares" it with ScamSlap and gets a verdict in 30 seconds. It's that simple.

🚀 Technical Arsenal (Under the Hood)

This system integrates cutting-edge technologies to ensure precision, speed, and security:

  • ⚡ FastAPI & Render: An ultra-fast Python backend hosted in the Cloud (Render), standing guard 24/7.
  • 🧠 Sightengine API (GenAI Model): Our elite mercenaries for AI scanning. We use advanced models that analyze media frames for mathematical traces of synthetic generation.
  • 🚧 YT-DLP Battering Ram: A digital breach tool that bypasses Facebook's walls to extract raw video links (.mp4) from Reels and Shares for deep analysis.
  • 🌐 Meta Graph API (Locale Detection): We query the Meta database to identify the user's language in real-time.
  • 🔄 deep-translator: We translate the verdict into the user's language right before the "bullet" is fired, ensuring maximum impact.
  • 🛡️ Secure Architecture: All API keys are hidden in Environment Variables on Render—never hardcoded, never exposed.

🛠️ How It Works

  1. Interception: The user sends a photo or a Reel to the ScamSlap Facebook Page.
  2. Triage (Background Tasks): The FastAPI server sends an immediate response to Facebook (200 OK) and pushes the heavy lifting to a background worker (to prevent duplicate messages).
  3. Breaching the Vault: If it's a Reel, yt-dlp extracts the direct video file.
  4. Sniper Scanning: We send the media to Sightengine with tactical sampling parameters (e.g., one frame every 5 seconds, capped at 4 frames) to cover the whole clip while optimizing API credits.
  5. Translation & Verdict: We translate the AI probability score (e.g., 98% AI) into a clear message in the user's language and send it back to Messenger.

Built with ❤️ to protect those who raised us.

Contributors