AI Content Detector

AI Content Detector

Python AI content detector: perplexity analysis, burstiness scoring, vocabulary richness, statistical text analysis, and confidence-scored reports. All math from scratch. Zero dependencies.

Part of the AI Toolkit collection by [CodeVault](https://ai-toolkit.codevault.dev).

Features

• Perplexity estimation — Bigram language model measures text predictability
• Burstiness scoring — Sentence length variation analysis (AI text is suspiciously uniform)
• Vocabulary richness — Type-token ratio detects AI's characteristic word diversity
• Transition word density — AI overuses words like "Furthermore", "Additionally", "Moreover"
• Repetition detection — N-gram repetition patterns common in AI output
• Readability scoring — Flesch reading ease and syllable analysis
• Confidence reports — Weighted ensemble verdict with per-signal breakdown
• Batch analysis — Analyze entire directories of text files
• JSON export — Machine-readable reports for integration into workflows

Quick Start

# Run demo with AI and human text samples
python src/ai_content_detector.py --demo

# Analyze inline text
python src/ai_content_detector.py --text "Your text to analyze goes here..."

# Analyze a file
python src/ai_content_detector.py --file document.txt

# Analyze with JSON export
python src/ai_content_detector.py --file document.txt --export report.json

# Batch analyze a folder
python src/ai_content_detector.py --batch essays/ --export results.json

# Quick verdict only
python src/ai_content_detector.py --file document.txt --quiet

Project Structure

ai-content-detector/
├── README.md
├── LICENSE
├── src/
│   └── ai_content_detector.py    # Core engine (~470 lines)
└── examples/
    ├── basic_usage.py             # Programmatic usage example
    └── sample_texts/              # AI and human text samples
        ├── ai_generated.txt
        └── human_written.txt

CLI Reference

... continues with setup instructions, usage examples, and more.