HumanMark

Open source AI content detection. One API. Zero dependencies.

---

Detect AI-generated text, images, audio, and video—entirely offline, with no external API calls.

Why HumanMark?

Feature	HumanMark	GPTZero, etc.
Self-hosted	✅	❌
Works offline	✅	❌
Zero cost	✅	❌
Privacy-first	✅	❌
Open source	✅	❌
Multi-modal	✅ Text, Image, Audio, Video	⚠️ Mostly text-only

Use Cases

For Developers

Use Case	Description
Content platforms	Filter AI spam from submissions
Hiring tools	Check candidate writing samples
Education apps	Verify student work
Social apps	Flag synthetic profiles/posts
CMS plugins	WordPress, Ghost, etc.
CI/CD pipelines	Lint content like you lint code
Browser extensions	Detect AI on any page

For Organizations

Who	Why HumanMark
Enterprises	Compliance requirements—can't send data to third-party APIs (HIPAA, GDPR, SOC2)
Government & Defense	Air-gapped networks, classified environments
Law firms	Attorney-client privilege, can't upload client docs to cloud
Healthcare	Patient data must stay on-premise
Banks & Financial	Regulatory restrictions on data sharing
Universities	High volume (100K+ students), budget constraints
Publishers & Media	Own the tool, don't rent it

Quick Start

Option 1: Go Install

go install github.com/vinpatel/humanmark/cmd/api@latest

Option 2: Docker

docker run -p 8080:8080 humanmark/humanmark

Option 3: Build from Source

git clone https://github.com/vinpatel/humanmark.git
cd humanmark
go run cmd/api/main.go

Server runs at http://localhost:8080

Usage

Detect Text

curl -X POST http://localhost:8080/verify \
  -H "Content-Type: application/json" \
  -d '{"text": "Your content here"}'

Response

{
  "id": "abc123",
  "human": true,
  "confidence": 0.85,
  "content_type": "text"
}

Detailed Analysis

curl -X POST "http://localhost:8080/verify?detailed=true" \
  -H "Content-Type: application/json" \
  -d '{"text": "Your content here"}'

How It Works

HumanMark uses statistical and forensic analysis—no ML models required.

Text Detection

Signal	Human	AI
Sentence length variance	High	Low
Vocabulary richness	Varied	"Safe" words
Contractions	"don't", "I'm"	"do not", "I am"
Punctuation variety	!?;:—	Mostly periods
AI phrases	Rare	"As an AI...", "It's important to note..."

Image Detection

Signal	Real Photo	AI Image
EXIF metadata	Present	Missing/fake
Camera make	Apple, Canon, etc.	None
Sensor noise	Natural pattern	Too clean

Audio and Video Detection

Analyzes format metadata, encoder signatures, and AI tool markers.

API Reference

Endpoint	Method	Description
/verify	POST	Analyze content
/verify/{id}	GET	Get result by ID
/health	GET	Health check

Configuration

Variable	Default	Description
PORT	8080	Server port
ENV	development	Environment
LOG_LEVEL	info	Logging level

Contributing

We welcome contributions! See CONTRIBUTING.md.

Help us improve accuracy by reporting false positives/negatives.

License

MIT License. See LICENSE.

---

Star this repo if you find it useful!