Inspiration

Cyberbullying affects 1 in 3 children worldwide. Most platforms only act after harm is done. We wanted to stop it at the source — the moment a harmful message is being typed.

What it does

EmoShield uses Claude AI to analyze messages in real-time before they are sent. It detects toxic emotions, gives a 0-100 safety score, blocks harmful messages, generates a kinder alternative, and delivers voice alerts via ElevenLabs in 6 languages.

How we built it

Claude AI (claude-sonnet-4) for real emotion analysis, ElevenLabs for voice alerts, Web Speech API as fallback, pure HTML/CSS/JavaScript — runs in any browser instantly with zero installation.

Challenges

Getting Claude AI to return consistent analysis for every type of message across multiple languages, and making the UI feel emotional and real-time.

What's next

Browser extension for WhatsApp and Instagram, school dashboard for parents and teachers, full ElevenLabs API voice integration.

Built With

Share this project:

Updates