Scam Detection
Uses intelligent analysis to estimate whether messages, links, documents or interactions may be malicious.
AI for Social Protection
GUARDIAN AI
Artificial intelligence for digital scam detection and prevention. GUARDIAN AI combines cybersecurity intelligence, machine learning, document inspection and evidence correlation to help users identify suspicious digital content.
Cyber Intelligence
Combines URL analysis, certificate validation, DNS inspection and website trust indicators.
Risk Assessment
Presents understandable reasons behind each detected risk and confidence score.
Digital Protection
Designed to support cybersecurity awareness, fraud prevention and safer online decisions.
Overview
What is GUARDIAN AI?
GUARDIAN AI is a cybersecurity and artificial intelligence project focused on detecting, analyzing and preventing digital scams through automated evidence analysis.
The platform combines machine learning, document inspection, website intelligence and fraud detection techniques to identify suspicious digital content and estimate the likelihood that a message, website, document or interaction is malicious.
The project is being developed as a research initiative focused on applying artificial intelligence to social protection, cybersecurity awareness and digital fraud prevention.
Motivation
Why this project matters
Digital scams have become increasingly sophisticated and accessible to attackers. Many people are exposed daily to phishing emails, fraudulent websites, fake invoices, social engineering attacks, smishing, messaging scams, cryptocurrency fraud and identity theft attempts.
Many victims lack the technical knowledge required to identify these threats before damage occurs.
GUARDIAN AI aims to bridge this gap by providing an intelligent assistant capable of evaluating multiple risk indicators and presenting understandable security recommendations.
Research Question
Main investigation
Can AI improve digital scam identification?
The central research question is whether artificial intelligence combined with cybersecurity intelligence can improve the identification of digital scams while remaining accessible to non-technical users.
Objectives
What GUARDIAN AI aims to achieve
Detect phishing attempts and social engineering patterns.
Analyze suspicious URLs and evaluate website trustworthiness.
Inspect PDF, DOCX, XLSX and ZIP files for suspicious indicators.
Correlate evidence from messages, links, domains and documents.
Reduce false positives through multi-signal analysis.
Provide explainable risk assessments for non-technical users.
Assist users in making safer decisions online.
Core Features
Analysis engines
Message Analysis
Analyzes emails, SMS messages, chat conversations, social media messages and support communications for urgency language, credential requests, financial pressure and social engineering.
Link Analysis
Evaluates URLs and websites through SSL/TLS certificate validation, domain reputation, domain age, DNS inspection, redirect chains, typosquatting and brand impersonation signals.
Document Analysis
Inspects PDF, DOCX, XLSX and ZIP files for embedded URLs, suspicious macros, obfuscated content, executable payload indicators and potential exploit vectors.
Evidence Correlation
Combines multiple observations instead of relying on a single indicator, generating a final confidence score based on the full risk context.
Explainable AI
Explains why a decision was made, which indicators were detected, what evidence contributed to the score and how confident the system is in its assessment.
User Report
Produces understandable security recommendations so users can make safer decisions before clicking, opening or sharing content.
Technical Stack
Technologies and concepts
Python
Flask
PostgreSQL
Docker
HTML
CSS
JavaScript
Machine Learning
XGBoost
LightGBM
Random Forest
Ensemble Learning
URL Intelligence
Certificate Validation
DNS Analysis
Threat Intelligence
Architecture
System organization
GUARDIAN AI/
├── User Input
├── Data Processing
├── Message Engine
│ ├── Urgency Detection
│ ├── Credential Request Detection
│ ├── Financial Pressure Signals
│ └── Social Engineering Patterns
├── Link Engine
│ ├── SSL/TLS Certificate Validation
│ ├── Domain Reputation
│ ├── DNS Inspection
│ ├── Redirect Analysis
│ └── Typosquatting Detection
├── Document Engine
│ ├── Embedded URL Extraction
│ ├── Macro Indicators
│ ├── Obfuscated Content Detection
│ └── Exploit Vector Signals
├── Evidence Correlation Layer
├── Machine Learning Engine
├── Risk Assessment
└── User ReportExample Output
How results could be presented
Scam Probability: 87%
Indicators:
✓ Urgency tactics detected
✓ Credential request detected
✓ Suspicious sender pattern
✓ Financial pressure language
Risk Level: High
Reasons:
- Domain registered recently
- Similar to a known banking website
- Multiple redirects detectedSocial Impact
Who can benefit?
GUARDIAN AI aims to contribute to digital inclusion, cybersecurity education, fraud prevention, online safety awareness and protection of vulnerable populations.
Potential beneficiaries include students, elderly users, small businesses, educational institutions and public organizations.
Future Components
Planned features
Browser extension for real-time scam and phishing detection.
Email plugin for analyzing suspicious messages.
SMS protection module for smishing detection.
Mobile application for safer digital interactions.
Community reporting system for suspicious content.
Real-time threat intelligence integration.
Explainable AI dashboard for transparent risk analysis.
Multi-language support for broader accessibility.
Research Status
Current phase
GUARDIAN AI is currently under active research and development. The project is being prepared for scientific presentation and evaluation, with a focus on cybersecurity education, social protection and explainable artificial intelligence.
The repository may remain private during the research and evaluation phase and can be released publicly after the project presentation and scientific publication process.
Trust, but verify.
Explore GUARDIAN AI, a research project focused on intelligence for safer digital interactions.
Open GitHub Repository