🔐 AI API Key Leak Detection · Leak Radar

Paste your OpenAI / Anthropic / Google AI / Groq / Perplexity / HuggingFace / xAI API Key. After browser-side SHA-256 hashing, it is matched against a leak fingerprint database continuously collected from the public internet. Your raw key never leaves this device.

Privacy architecture: window.crypto.subtle.digest("SHA-256") runs entirely in your browser. The server only receives the 64-character hexadecimal hash — your raw key never leaves this device. For learning and research use only.

📡 Public Fingerprint Database — Live Stats

145Verified leak fingerprints
0New in 24h
7Active sources
9Vendors affected
150Total candidates

Data updated at 2026-07-30 16:28 UTC · cron collects incrementally every 2 hours

By Vendor — Whose Keys Leak Most

VendorVerified leaksShare
Google AI Studio 80 55.2%
Groq 21 14.5%
OpenAI Project 17 11.7%
OpenRouter 7 4.8%
Anthropic API03 5 3.4%
DeepSeek 5 3.4%
Perplexity 4 2.8%
HuggingFace 3 2.1%

By Source — Where They Were Found

SourceVerified leaksShare
github_code_search 42 29.0%
huggingface 39 26.9%
pypi 37 25.5%
postman 12 8.3%
deploy_js 9 6.2%
github_gist 4 2.8%

How This Project Works

📊 Collection: Continuous passive scanning and deduplication from 10 public sources (public code repositories, public package indexes, public deployment artifacts, etc.). Only publicly accessible content is fingerprinted.
🔒 Storage: The fingerprint database stores only hashes and source metadata — raw keys never enter the main table. Hashing is done locally in your browser; your raw key never leaves this device.
🔬 Purpose: For learning and research only — helps developers quickly check whether their keys have appeared on the public internet. No validity verification is performed; no upstream requests are made.

📘 Full methodology + privacy architecture + source details →