Privacy Mask
A free tool that automatically blurs out phone numbers, ID cards, and API keys in a screenshot before you paste it into an AI chat, so you don't accidentally leak personal data.
🔗 Visit Privacy Mask
Description
Pasting a screenshot into ChatGPT or Claude to ask 'what's wrong with this UI' or 'help me debug this' is a common habit — but that screenshot often has a phone number, an ID card, or a stray API key visible in it, and once it's sent, it's sent. Privacy Mask catches that before it happens: it scans a screenshot locally, finds anything sensitive, and blurs or blocks it out automatically.
Privacy Mask is a free, open-source, MIT-licensed tool that detects over 40 categories of sensitive data — phone numbers, national ID card formats from 15+ countries, financial data, and API keys — using a combination of named-entity recognition and regex pattern matching, backed by dual OCR engines (Tesseract and RapidOCR) to read text embedded in images. Everything happens 100% locally and offline, so the original unredacted screenshot never has to leave your machine, and it integrates with 20+ AI coding platforms so it can sit in your existing workflow rather than requiring a separate manual step.
💬 Our review
The short version: Privacy Mask solves a real, easy-to-overlook leak point — the screenshot you paste into an AI chat without checking what's actually visible in it — and does it entirely offline, which matters given the sensitive nature of what it's scanning.
The local-only processing is the right design choice here: a tool that redacts sensitive data but has to upload the original image to do so would defeat its own purpose. Combining dual OCR with NER and regex detection across 40+ pattern types and 15+ national ID formats is also broader coverage than a simple regex-only redactor would offer.
The honest gaps: it's a small, early project (11 stars, 1 fork) from a lesser-known author, so it hasn't had the scrutiny a widely-used privacy tool ideally gets, and automated detection of this kind is never perfect — a determined check of the output before sharing is still worth doing. For anyone who regularly shares screenshots with AI tools, it's a low-cost, sensible safety net; it shouldn't be treated as a guaranteed leak-proof filter.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Entièrement gratuit et open-source (MIT), traitement 100% local et hors-ligne.
Pros
Traitement 100% local — l'image originale ne quitte jamais la machine
Détection de 40+ types de données sensibles (téléphones, cartes d'identité, clés API)
Double moteur OCR (Tesseract + RapidOCR) pour le texte intégré dans l'image
Intégration avec 20+ plateformes IA de code
Cons
Projet jeune et peu connu (11 étoiles, 1 fork)
La détection automatique n'est jamais garantie à 100%
Nécessite une vérification manuelle pour les cas critiques