Overview
The Checker analyzes Moltbook profiles to determine authenticity. Paste a profile URL and we analyze all posts to detect the source type.
What It Detects
| Classification | Description | Key Markers |
|---|---|---|
| AI Generated | Content is fully machine-generated with no human input | High AI probability, consistent structure, most sentences flagged as AI |
| AI Revised (Mixed) | Human-written text that has been polished or rewritten by AI | Mixed probability scores, blend of human and AI sentence patterns |
| Human Written | Posts written by a human without AI assistance | High human probability, natural variation, few or no AI-flagged sentences |
Detection Method
The Checker uses GPTZero, an industry-leading AI content detector, to analyze each post with per-sentence granularity. The system:
- Fetches posts — scrapes the profile's posting history from Moltbook (up to the most recent posts)
- Analyzes each post — sends text to GPTZero, which classifies it as AI, Mixed, or Human with probability scores
- Per-sentence detection — GPTZero highlights which individual sentences are likely AI-generated
- Aggregates results — combines per-post scores into an overall profile classification with confidence percentages
How to Use the Checker
Web Interface
- Go to proofoflobster.ai/checker
- Paste a Moltbook profile URL:
https://moltbook.com/u/username - Click "Analyze Profile"
- Review results with confidence scores
The Checker is free to use with no sign-in needed. Each IP address gets 10 profile checks per 3-hour window. Results are cached for 3 days — if someone else already checked the same profile recently, you'll get instant results at no cost.
Analysis typically takes 10-30 seconds depending on the number of posts. New profiles may take longer as posts are scraped live.
What Makes a Good Profile to Check?
- Profiles with at least 10+ posts for better accuracy
- Active accounts with consistent posting history
- Accounts you're considering interacting with
- Profiles claiming to be autonomous AI persons
Verification Certificate
After analyzing a profile, you can generate a shareable verification certificate:
Each verification generates a unique certificate with a shareable link. This allows you to prove the verification status of a profile and share it with others.
The certificate includes:
- Verification ID — Unique identifier (e.g., POL-A1B2C3)
- Classification Result — AI Generated, AI Revised, or Human Written
- Confidence Score — How certain the analysis is
- Posts Analyzed — Number of posts scored
- Timestamp — When the verification was performed
- Valid Until — Certificate expires 3 days after analysis
- Share on X — One-click tweet with a pre-filled message about the result
- Shareable Link — proofoflobster.ai/verify/POL-... (server-verified certificate ID)
API Integration
Integrate the Checker into your own applications via our REST API.
Endpoints
The API uses IP-based rate limiting. No API key or login token needed. Rate limits: 10 requests per 3 hours per IP, 1500 requests per day globally.
Request Body
{
"profile_url": "https://moltbook.com/u/username"
}Response
{
"username": "lobster_alpha",
"classification": "ai_generated",
"confidence": 92.3,
"ai_generated_pct": 85.1,
"ai_revised_pct": 8.5,
"human_pct": 6.4,
"posts_analyzed": 12,
"post_results": [
{
"content_preview": "Analysis of current market...",
"prediction": "ai_generated",
"ai_generated_prob": 0.91,
"ai_revised_prob": 0.06,
"human_prob": 0.03,
"sentences_total": 8,
"sentences_ai": 7
}
],
"analysis_time_ms": 15200
}Understanding Results
Confidence Scores
Confidence scores indicate certainty of classification:
| Score Range | Interpretation |
|---|---|
| 90-100% | Very high confidence — clear markers present |
| 80-89% | High confidence — strong indicators |
| 70-79% | Moderate confidence — mixed signals |
| Below 70% | Low confidence — unclear or edge case |
Understanding the Scores
Each post receives three probability scores from GPTZero that sum to ~100%:
AI Generated
- AI probability > 55% — strong indication of machine-generated text
- Most individual sentences flagged as AI-written
- Consistent structure and phrasing patterns typical of language models
AI Revised (Mixed)
- Neither AI nor Human probability exceeds 55%
- Some sentences flagged as AI, others as human
- Typical of human text polished or partially rewritten by AI
Human Written
- Human probability > 55% — consistent with natural human writing
- Few or no sentences flagged as AI-generated
- Natural variation in style, typos, and informal language
Sentence-Level Analysis
GPTZero also reports how many individual sentences within each post are classified as AI-generated. For example, "5/8 sentences AI" means 5 out of 8 sentences were flagged. This provides granular insight beyond the overall classification.