FOSS Hack 2025 > All Projects > Telegram CryBIT

Telegram CryBIT

Telegram CryBIT is an advanced, AI-powered scam detection system designed to monitor and analyze messages on Telegram in real-time.

Description

Telegram CryBIT is an advanced, AI-powered scam detection system designed to monitor and analyze messages on Telegram in real-time. The system leverages Machine Learning (ML), Natural Language Processing (NLP), and pattern-based techniques to detect and flag fraudulent content, such as cryptocurrency scams, phishing links, and fake investment schemes. CryBIT is built using Flask for the backend, Telethon for Telegram integration, and MongoDB for data storage. It features a real-time risk scoring system, an interactive UI, and dynamic configuration options, making it a powerful tool for preventing financial fraud.

The core of CryBIT lies in its multi-layered scam detection logic, which combines ML-based text classificationNLP semantic similarity matchingkeyword-based detectionphishing URL detectioncrypto wallet blacklist checks, and OCR-based text extraction. Each detection technique contributes to a cumulative risk score, which is compared against a configurable threshold to determine if a message is a scam. If the risk score exceeds the threshold, the message is flagged, stored in MongoDB, and displayed in the dashboard for further analysis. Admins are also alerted in real-time via Telegram.

CryBIT's scoring system is designed to be flexible and accurate. Each detection method adds a specific weight to the risk score:

  • ML Model Prediction: Adds +0.6 to the score.

  • Keyword Match: Adds +0.3.

  • Phishing URL Detected: Adds +0.6.

  • Crypto Wallet Blacklist Match: Adds +0.7.

  • OCR Text Match: Adds +0.3.

The total risk score is calculated as the sum of these weights. If the score exceeds the risk threshold (default: 0.5), the message is flagged as a scam. This multi-layered approach ensures high accuracy and minimizes false positives.

CryBIT's modular architecture ensures scalability and ease of maintenance. The system is divided into several modules, including configuration managementutility functionsFlask backendscam detection logicTelegram monitoring, and database integration. The interactive UI allows users to manage monitored channels, view flagged messages, and customize scam detection settings.

Issues & PRs Linked
No Issues, PRs or Discussions added.