TRAST Documentation
TRAST.spaceConnect
  • Introduction
    • What is TRAST?
    • TRAST Litepaper
    • Trust System
    • Mechanics
    • Core Concepts
  • Getting Started
    • Quick Start Guide
    • Beta Information
    • Understanding Trust Scores
  • Platform Features
    • Platform Overview
    • Core Infrastructure
      • Vote Staking
      • Trust & Reviews
      • TRAST Premium
      • Verification
      • Feature Specifications
      • Tokenomics Analysis (Planned)
    • PayAI & Marketplace
      • PayAI System
      • TRAST Marketplace
      • TRAST.fun (Planned)
      • Advanced Platform Features
      • Bridge to XRPL
    • User Interaction
      • Inline Search
      • Live Feed
      • Commenting System
      • Identity Integration
      • Smart Onboarding
      • Monetization Options
    • Chat System
      • Chat Rooms
      • Specialized Rooms
      • Two-Tier Rooms
    • Platform Analytics
  • Technical Documentation
    • Technical Overview
    • Core Architecture
      • Design & Implementation
      • Security & Trust
      • Infrastructure Costs
    • AI Systems
      • Bot System
      • AI Chat Guide
      • AI Scam Radar
      • AI Review Copilot & TrustDigest
      • Content Agent
    • Advanced Features
      • Bot Outputs
      • Customization & Integrations
  • Community
    • Guidelines
      • Communication Style
      • Communication Playbook
      • Community Features
    • Engagement & Rewards
      • Community Incentives
      • Premium Features (Gem Hunters)
      • TRAST Reflections (Blog)
  • Resources
    • Use Cases
      • Overview
      • Detailed Examples
      • Real World Scenarios
    • Support & Contact
      • Contact Information & Company Details
      • Support Center
  • Token
    • Economics
      • Financial Model
Powered by GitBook
On this page
  • How Scam Radar Works
  • Risk Score & Levels
  • Output & Usage
  • Limitations & Future Development
  1. Technical Documentation
  2. AI Systems

AI Scam Radar

Learn about TRAST's Scam Radar, an AI-powered system combining LLM analysis, on-chain heuristics, and social signals to detect risky Web3 entities.

TRAST's Scam Radar™ is a sophisticated AI-powered system designed to provide real-time risk assessment for Web3 entities. It proactively identifies potential scams and high-risk projects by analyzing multiple data points.

How Scam Radar Works

The system combines several analysis methods to generate a comprehensive risk score (0-100) and actionable insights:

  1. LLM-Based Risk Analysis (Weight: 40%):

    • Uses advanced language models (OpenAI via LLMService) to analyze project descriptions, websites, and marketing materials.

    • Detects common red flags: unrealistic promises, anonymous teams, vague language, excessive hype, lack of documentation, etc.

  2. On-Chain Heuristics (Weight: 30%) - Future Implementation:

    • Currently placeholder. Will analyze blockchain transaction patterns, token distribution, smart contract interactions, and holder behavior to identify suspicious on-chain activity (e.g., wash trading, rug pull indicators).

  3. Social Signal Analysis (Weight: 15%) - Future Implementation:

    • Currently placeholder. Will evaluate social media presence (Twitter, Telegram, Discord), community engagement levels, sentiment analysis, and follower authenticity to detect manufactured hype or lack of genuine community.

  4. Code Scanning (Weight: 15%) - Future Implementation:

    • Currently placeholder. Will integrate with tools like Slither or MythX to analyze smart contract source code for known vulnerabilities, malicious functions, or unsafe patterns.

Risk Score & Levels

The individual scores from each analysis method are combined using the weights above to produce a final risk score. This score is then mapped to a clear risk level:

  • Low: 0-30

  • Moderate: 31-60

  • High: 61-85

  • Critical: 86-100

Output & Usage

  • Scam Radar results include the overall score, risk level, and a list of specific reasons (red flags) identified during the analysis.

  • This information is integrated into entity profiles and search results within TRAST.

  • Users receive warnings for high-risk entities, enabling informed decision-making.

Limitations & Future Development

  • Currently, the score relies heavily on LLM analysis as on-chain, social, and code analyses are under development.

  • The system is continuously learning and improving its detection capabilities.

  • Future versions will incorporate more data sources and refine the weighting algorithm.

Scam Radar provides a crucial layer of automated protection, but always remember to conduct your own thorough research (DYOR) before engaging with any Web3 entity.

PreviousAI Chat GuideNextAI Review Copilot & TrustDigest

Last updated 11 days ago