🧠 AI-Content Detector ML
Automatic detection of ChatGPT, Claude, Gemini & Co. with 15,234 training samples and 92% accuracy.
0
Active Users
0
Detected AI Texts
92%
Accuracy
15.234
Training Samples
🚀 Features
What makes our detector special
Machine Learning
Trained with 15,234 real samples from humans and various AI models for highest precision.
Dark Mode
Automatic Dark Mode support - perfect for nighttime browsing. Adapts to your system.
6 AI Models detectable
Detects GPT-3, GPT-4, Claude, Gemini, Llama and other AI models with color-coded badges.
Blur Effect
AI content is gently blurred. Click to reveal and read anyway.
ML Status Display
Live display of ML status and detection confidence directly on the webpage.
Dynamic Phrase Loader
Over 900 AI phrases in German, English and Vietnamese loaded on demand.
🎮 Interactive Demo
Test the detection live
⬇️ Download
For all major browsers
🔧 System Requirements: Chrome 88+, Firefox 109+, Edge 90+, Opera 75+, Brave
📦 Size: ~25MB (after training)
v3.1 ML❓ Frequently Asked Questions
How accurate is the detector? ▼
Our ML model achieves about 92% accuracy on our test data. Accuracy may vary depending on text length, language, and complexity. For very short texts (<50 words), we recommend viewing results critically.
Does the plugin work offline? ▼
Yes! After the initial ML model download (~25MB), the plugin works completely offline. All analyses run locally in your browser - your data never leaves your computer.
Which languages are supported? ▼
The detector supports German, English, and Vietnamese. Language detection is automatic. For other languages, accuracy may be lower as our training data mainly covers these three languages.
Is the plugin free? ▼
Yes, the AI-Content Detector ML is and remains completely free. We believe in transparent and accessible technology for everyone.
Is my data stored? ▼
No! All analyses run locally in your browser. No data is transmitted to our servers or third parties. The only stored data are your settings and whitelist (locally in your browser).
How can I report false detections? ▼
Simply use our contact form or send us an email. We continuously improve our model with real user feedback.
What is the minimum text length for detection? ▼
For reliable results, we recommend texts with at least 50 words. Shorter texts can be analyzed but the confidence level may be lower.
Does it work with social media posts? ▼
Yes, the detector works with any text-based content including social media posts, articles, comments, and emails.
🔧 Technical Details
Framework
TensorFlow.js
Training Data
15.234
Features
19
Epochs
20
Target Accuracy
92%
Backend
WebGL/CPU