This contains everything you need to run your app locally.
View your app in AI Studio: https://ai.studio/apps/drive/1peWiryxF8gqDpD8wlgXouy2AgYnZC0EQ
Prerequisites: Node.js
- Install dependencies:
npm install - Set the
GEMINI_API_KEYin .env.local to your Gemini API key - Run the app:
npm run dev
Saif-AI is a high-performance, Gemini-powered cybersecurity agent specifically engineered to defend UAE residents from digital fraud through advanced multimodal analysis of suspicious messages and screenshots. Serving as a comprehensive digital shield, the application allows users to instantly scan phishing attempts in five major languages—English, Arabic, Hindi, Urdu, and Tagalog—with specialized detection modules for prevalent logistics and government-ID scams. Beyond its core diagnostic engine, Saif-AI provides real-time community protection through a live threat heatmap integrated with Google Maps, an automated forensic reporting system for the Dubai Police Al Ameen service, and gamified educational tools designed to build natural cyber-immunity across the Emirates' diverse population.
Future Direction: We will further implement this in androids as well as windows with messaging service and emails to send an automated response to users if scam messages or emails are detected.
🛡️ Saif-AI: UAE Digital Defense Core
Saif-AI (Arabic for "Sword") is a next-generation cybersecurity agent designed specifically for the United Arab Emirates. Powered by Google Gemini 3 Flash and Lingo.dev, it acts as a multilingual shield against phishing, courier fraud, and identity theft targeting UAE residents.
Cybercrime in the UAE is evolving. Residents face sophisticated scams including:
- Logistics Fraud: Fake SMS from Emirates Post, Aramex, or DHL demanding "customs fees."
- Government Impersonation: Phishing sites mimicking UAE Pass or Dubai Police.
- Language Barriers: Vulnerable demographics are often targeted in languages they don't speak fluently.
Saif-AI is a web-based "Cyber Defense Core" that allows users to scan text and screenshots instantly. It doesn't just say "Scam" or "Safe"—it explains why, translates the analysis into 5 languages, and helps users report the threat to authorities immediately.
- Image Analysis: Users can upload screenshots of suspicious WhatsApp messages or emails. The AI performs OCR and visual analysis to detect forged logos and urgency cues.
- Text Analysis: Paste raw text or URLs for instant verification.
- Risk Scoring: Returns a 0-100 risk score with a tactical breakdown of the threat.
To serve the UAE's diverse population, the app helps verify threats in:
- 🇬🇧 English
- 🇦🇪 Arabic
- 🇮🇳 Hindi
- 🇵🇰 Urdu
- 🇵🇭 Tagalog
- Powered by the Lingo.dev API for high-accuracy, context-aware translation.
A specialized module trained to recognize logistics fraud patterns. It specifically looks for tracking URL spoofing (e.g., emirates-post.net vs emiratespost.ae) and "small fee" payment gateways.
A real-time visualization overlay (using Google Maps) that displays a heat map of active scam reports across Dubai, Abu Dhabi, and Sharjah, keeping the community informed.
If a scam is detected, Saif-AI auto-generates a pre-formatted SMS report for Al Ameen Service (8004888), allowing users to report threats to UAE Law Enforcement in seconds.
An interactive "Real vs. Scam" quiz module that awards XP and ranks (Novice to Cyber Guardian), helping users build natural defensive instincts.
| Component | Technology | Description |
|---|---|---|
| Frontend | React 19 | Latest React features for a responsive UI. |
| Styling | Tailwind CSS | "Cyberpunk" dark mode aesthetic with glassmorphism. |
| AI Brain | Google GenAI SDK | Uses gemini-3-flash-preview for ultra-fast multimodal analysis. |
| Localization | Lingo.dev API | Handles dynamic translation of UI and Analysis results. |
| Icons | Lucide React | Lightweight, consistent iconography. |
| Maps | Google Maps Embed | Visualizing threat data. |
- Node.js (v18 or higher)
- Google Gemini API Key:
- Lingo.dev API Key
-
Clone the repository:
git clone https://github.com/yourusername/saif-ai.git cd saif-ai -
Install dependencies:
npm install
-
Configure Environment: Create a
.envfile in the root directory:# Your Google Gemini API Key API_KEY=your_google_api_key_here # (Optional) Key is currently hardcoded in service for demo purposes, # but best practice is to put Lingo key here too.
-
Run the Development Server:
npm run dev
-
Open your browser at
http://localhost:5173.
- The "Customs Fee" SMS: A user receives an SMS asking for 15 AED for a package. They upload a screenshot to Saif-AI. The AI detects the fake URL and warns the user it is a scam.
- The "Police Fine" Email: A user gets a PDF claiming to be a fine. They paste the text. Saif-AI cross-references the official Dubai Police email format and flags it as phishing.
- Non-English Speakers: An Urdu-speaking resident uses the app in Urdu to understand why a "Bank Account Frozen" message is fake.
- Voice Mode: Real-time audio analysis for incoming phone calls to detect AI voice cloning and vishing (Voice Phishing).
- WhatsApp Bot: Integration directly into WhatsApp for easier forwarding of suspicious messages.
- Community API: An open API for local banks to feed real-time scam data into the Threat Map.
Saif-AI is a tool for analysis and education. While it uses advanced AI to detect patterns, it does not guarantee 100% accuracy. Always contact official entities (Bank, Police, Post Office) directly through official channels to verify claims. In case of financial loss, contact UAE authorities immediately.
Made with ❤️ in Dubai.