An AI-powered global guardian that protects people—especially elders—from online scams using empathy, language, cultural awareness, and intelligence.
Kavach AI is a high-impact Streamlit web application built for global social good. It analyzes suspicious messages and screenshots to help users identify online scams—without using technical jargon—and adapts its communication based on the user’s country, language, and cultural context.
Instead of sounding like a cybersecurity tool, Kavach AI behaves like a trusted family member or personal guardian, guiding users calmly and clearly through risky digital situations.
Online scams are a global problem affecting millions every day. People fall victim not because they lack intelligence, but because:
- Scam messages create fear and urgency
- Technical explanations are confusing
- Help is not immediate, personal, or localized
Kavach AI solves this by combining AI intelligence with human empathy and cultural awareness.
- Country / region selector
- Scam understanding adapts to regional fraud patterns
- Responses generated using native language, tone, and cultural norms
- Provides local cybercrime reporting guidance based on location
- Text Analysis: Paste SMS, WhatsApp, Email, or social media messages
- Image Analysis (OCR + Vision): Upload screenshots (JPG / PNG / JPEG)
- Detects urgency, impersonation, fake branding, and malicious intent
- Multilingual support with culturally appropriate phrasing
- Avoids technical terms like phishing, malware, or HTTPS
- Speaks gently, clearly, and reassuringly
“Please don’t click this link. It’s trying to scare you so you act fast. I’m here to keep you safe.”
- Designed for text + voice output
- Ideal for elders and non-technical users
- Ready for future text-to-speech integration
- 🟢 Safe
- 🟡 Suspicious
- 🔴 Dangerous
Every analysis ends with exactly 3 simple steps, such as:
- Do not click or reply
- Block the sender
- Report to your local cybercrime authority
| Text Analysis | Image Upload | Risk Result |
|---|---|---|
| Layer | Technology |
|---|---|
| Frontend | Streamlit (Python) |
| AI Engine | Google Gemini 1.5 Flash |
| Multimodal Processing | Gemini Vision + OCR |
| Localization Engine | Country & Language Mapping |
| Styling | Custom CSS |
| Deployment | Streamlit Cloud / Cloud VM |
flowchart TB
U[End User]
U --> UI[Streamlit Frontend]
UI --> TIN[Text Input]
UI --> IIN[Image Upload]
UI --> LOC[Country Selector]
UI --> LANG[Language Selector]
TIN --> IP[Input Processing]
IIN --> IP
IP --> TV[Text Validation]
IP --> IV[Image Validation]
IV --> ERR[Error Handling]
TV --> CTX[Localization Engine]
IV --> CTX
LOC --> CTX
LANG --> CTX
CTX --> PROMPT[Prompt Engineering]
PROMPT --> AI[Gemini 1.5 Flash API]
AI --> RISK[Risk Classification]
RISK --> SAFE[Safe]
RISK --> SUSP[Suspicious]
RISK --> DANG[Dangerous]
SAFE --> RESP[Response Formatter]
SUSP --> RESP
DANG --> RESP
RESP --> OUT[Localized Output]
OUT --> UI
- API keys are masked and stored only during runtime
- No user messages or images are stored
- All analysis is performed in real time
# Clone the repository
git clone https://github.com/your-username/kavach-ai.git
cd kavach-ai
# Install dependencies
pip install -r requirements.txt
# Run the app
streamlit run app.py
Fake bank or payment alerts
Urgent delivery or courier scams
Government or authority impersonation
Lottery, prize, or reward fraud
Social engineering messages on messaging apps
Protects users across countries, cultures, and languages
Makes cybersecurity understandable, not intimidating
Reduces fear-driven digital fraud
Encourages timely reporting and awareness
“Security should feel human—everywhere in the world. Because trust isn't built on code; it's built on understanding.”
Voice input & text-to-speech output
WhatsApp / Messenger bot integration
Country-specific scam trend dashboard
Family or caregiver alert system
Offline-first lightweight mode
- Plug-in support for:
- Voice input / output
- Messaging platform bots
- Country-specific scam datasets
- Can be deployed on:
- Streamlit Cloud
- Cloud VM
- Containerized environments
Kavach AI is not just a detection system.
It is a communication system built on empathy, localization, and trust.
Kavach AI is built with the belief that technology should protect people, not overwhelm them.
If you find this project valuable, ⭐ the repository and help spread digital safety awareness.
This project is licensed under the MIT License — free to use for innovation and social good.