Skip to content
manuscript.dev

Introduction

What is Manuscript?

Section titled “What is Manuscript?”

Manuscript is an open-source, privacy-first AI content detector that runs entirely on your infrastructure. It can detect AI-generated content across four modalities:

  • Text - Detect ChatGPT, Claude, Gemini, and other LLM-generated content
  • Images - Identify DALL-E, Midjourney, Stable Diffusion generated images
  • Audio - Spot ElevenLabs, Suno, and other AI voice/music generators
  • Video - Detect deepfakes and AI-generated video content

Why Manuscript?

Section titled “Why Manuscript?”

Every other AI detection service requires you to upload your content to their servers. That’s a dealbreaker for many organizations:

IndustryWhy On-Premise Matters
HealthcareHIPAA compliance prohibits external data sharing
LegalAttorney-client privilege can’t survive third-party uploads
FinanceSOC2/PCI requirements restrict data sharing
GovernmentAir-gapped networks, classified environments
EducationScale without per-seat licensing costs

Manuscript runs entirely on YOUR infrastructure. Your data never leaves your network.

How It Works

Section titled “How It Works”

Unlike ML-based detectors that require GPUs and large models, Manuscript uses statistical and forensic analysis:

Text Detection

Section titled “Text Detection”

Analyzes linguistic patterns that differentiate human and AI writing:

  • Sentence length variance (humans vary more)
  • Vocabulary richness and rare word usage
  • Contraction patterns (“don’t” vs “do not”)
  • Known AI phrases (“As an AI…”, “It’s important to note…”)
  • Hedging language and repetition patterns

Image Detection

Section titled “Image Detection”

Examines forensic signals in image files:

  • EXIF metadata presence and validity
  • Camera make/model signatures
  • Sensor noise characteristics
  • Compression artifact patterns
  • Color distribution analysis

Audio Detection

Section titled “Audio Detection”

Analyzes audio container and signal characteristics:

  • File header and container metadata
  • Encoding parameters and profiles
  • FFT spectral analysis
  • MFCC computation
  • AI tool fingerprints (ElevenLabs, Suno markers)

Video Detection

Section titled “Video Detection”

Combines multiple forensic approaches:

  • Container metadata analysis
  • Bitrate consistency checking
  • Temporal pattern analysis
  • Encoding signature detection

Key Features

Section titled “Key Features”
  • 100% Offline - No external API calls
  • Self-Hosted - Deploy on your infrastructure
  • Open Source - MIT licensed, fully auditable
  • Multi-Modal - Text, image, audio, video
  • Fast - Sub-10ms response times
  • Zero Dependencies - Pure Go implementation

Next Steps

Section titled “Next Steps”