# Occlira

> Occlira is a desktop app for Windows and macOS that detects and removes personal data (PII) from documents, spreadsheets, email, audio and images entirely on your own computer — no cloud, no account. You anonymize a file before sharing it or pasting it into an AI tool, then restore the original values locally when you are done.

Occlira runs 100% locally on Windows 10/11 and macOS. Detection uses a local AI model plus pattern rules, and audio is transcribed on-device; your files and the personal data inside them never leave your computer, and there is no telemetry by default. It supports 7 interface languages (English, German, French, Italian, Russian, Dutch, Spanish) and detects 40+ types of personal data across documents (DOCX, PDF, XLSX, EML, TXT, MD, CSV), audio (MP3, WAV, M4A, FLAC, OGG) and images (JPG, PNG, HEIC, WEBP, TIFF, BMP). For photos it can blur faces on-device and strip location/EXIF metadata. Anonymization is reversible: a mapping stored locally lets you restore the original values. It is a one-time license (from €149 for a single seat, up to €999 for a 10-seat team; the exact amount is shown at checkout) with a 14-day free trial, sold through Polar as Merchant of Record. A companion Chrome extension (included with the license, requires the desktop app) anonymizes text and attachments inside ChatGPT, Claude and Gemini.

## Core pages
- [Home / overview](https://occlira.com/): what Occlira is, how it works, use cases and pricing
- [Pricing](https://occlira.com/pricing/): one-time license (solo and team seats), 14-day free trial
- [Download](https://occlira.com/download/): the Windows and macOS installers

## Guides and answers
- [All guides](https://occlira.com/guides/): index of tasks, questions, professions and reference
- [Anonymize before ChatGPT](https://occlira.com/anonymize-before-chatgpt/): remove PII before pasting into an AI tool, then restore locally
- [Video: anonymize a Word document before ChatGPT (YouTube, 6 min)](https://www.youtube.com/watch?v=PQ-aNPhI2gA): the desktop workflow on screen — drop the files in, review what the model found, approve, restore the real names later; embedded in the guide above
- [Remove personal data from audio](https://occlira.com/redact-audio/): local transcription + redaction of recordings
- [Redact Word, PDF & Excel](https://occlira.com/redact-pii-word-pdf-excel/): remove PII from office files locally
- [Remove metadata from a Word document](https://occlira.com/remove-metadata-word-document/): Document Inspector on Windows, the Mac and web limits, the author name, and what it misses
- [Is ChatGPT private?](https://occlira.com/is-it-safe-to-use-chatgpt-with-confidential-data/): what OpenAI stores, who can see your chats, and the safe way to use it with confidential data
- [Does ChatGPT store your data?](https://occlira.com/does-chatgpt-store-your-data/)
- [Use AI without breaking the GDPR](https://occlira.com/use-ai-without-breaking-gdpr/)
- [What is PII? Types and examples](https://occlira.com/what-is-pii/)
- [Local vs cloud PII redaction](https://occlira.com/compare/local-vs-cloud-pii-redaction/)
- For law firms, healthcare & therapists, and accountants: https://occlira.com/for/law-firms/ · https://occlira.com/for/healthcare/ · https://occlira.com/for/accountants/
- [Blur faces & strip photo EXIF](https://occlira.com/blur-faces-exif/): cover faces with an opaque box and remove GPS/EXIF metadata locally (why blur is reversible)
- [Shadow AI at work](https://occlira.com/shadow-ai/): staff pasting company data into personal chatbots, and how to manage it
- [Is Claude safe?](https://occlira.com/is-claude-safe-for-confidential-data/): what Anthropic trains on and stores, who can see your chats, Cowork/Claude Code risk, and how Claude compares
- [Is Grok private? And is Grok safe to use?](https://occlira.com/is-grok-private/): what xAI stores on grok.com, in the apps and on X, the training opt-out, Private Chat, and the August 2025 shared-conversation exposure
- [Is DeepSeek safe to use?](https://occlira.com/is-deepseek-safe/): where the data is stored (People’s Republic of China), training and the opt-out, the government restrictions, and what running the open weights locally changes
- [For journalists](https://occlira.com/for/journalists/): protecting sources — strip metadata, redact for real, transcribe offline, blur faces
- [ABA Formal Opinion 512](https://occlira.com/aba-opinion-512/): the US ethics rules for lawyers using generative AI
- [Attorney-client privilege & AI](https://occlira.com/attorney-client-privilege-ai/): can using ChatGPT waive privilege?
- [PII detection benchmark](https://occlira.com/benchmark/): how detection quality is measured

## Legal & professional-secrecy guides (by jurisdiction)
- Germany — [§ 203 StGB and AI](https://occlira.com/de/203-stgb-ki/) (when professionals bound by secrecy risk criminal liability); [An AVV with OpenAI — and why it isn't enough](https://occlira.com/de/avv-openai-berufsgeheimnis/); [Client data in ChatGPT — what the BStBK FAQ and DATEV say](https://occlira.com/de/mandantendaten-chatgpt/)
- France — [AI and a lawyer's professional secrecy (the CNB guide)](https://occlira.com/fr/ia-secret-professionnel-avocat/); [The CNIL 2025 AI checklist](https://occlira.com/fr/cnil-ia-checklist/); [Article 226-13 of the Code pénal and AI](https://occlira.com/fr/article-226-13-ia/) (1 year / €15,000)
- Italy — [Law 132/2025 on AI and professional secrecy](https://occlira.com/it/legge-132-2025-ia/); [The CNDCEC operational AI guide for accountants](https://occlira.com/it/commercialisti-ia-cndcec/)

## Privacy and practices
- [Data & Privacy Practices](https://occlira.com/data-practices/): the local-first architecture, what is stored on your device and how to delete it, and the few network calls the app makes
- [Privacy Policy](https://occlira.com/privacy/): the limited data the vendor processes (license, support and website data only)
- [Terms of Use & License (EULA)](https://occlira.com/terms/): licensing, trial, warranties and governing law (Cyprus)

## Key facts
- Platform: Windows 10/11 and macOS
- Processing: 100% on-device; no cloud AI; no account; no telemetry by default
- What it detects: names, emails, phone numbers, addresses, IBANs and credit cards, national IDs (e.g. US SSN, UK NINO, EU VAT, passports, DE Steuer-ID, FR INSEE, IT Codice Fiscale, ES DNI), dates of birth, IP addresses, medical record numbers, and more
- Document formats: DOCX, PDF, XLSX, EML, TXT, MD, CSV
- Audio formats: MP3, WAV, M4A, FLAC, OGG
- Image formats: JPG, PNG, HEIC, WEBP, TIFF, BMP (face blur + EXIF/GPS metadata removal)
- Interface languages: English, German, French, Italian, Russian, Dutch, Spanish
- Reversible: a mapping stored locally lets you restore the original values after anonymizing
- Real PDF redaction: burns boxes that actually delete the underlying text and metadata (not just a black overlay)
- Pricing: one-time perpetual license from €149 (single seat) up to €999 (10 seats); 14-day free trial; sold via Polar (Merchant of Record)
- Two ways to work: (1) anonymize a file in the desktop app, use the placeholder copy with any AI (paste the text or upload the file), then Deanonymize the reply or processed file in one click to restore the real values locally; (2) three local connectors that need the desktop app running: the Chrome extension (Anonymize / Shield file / Restore inside ChatGPT, Claude and Gemini), the Claude Desktop connector (MCP; only file paths are shared, real data never leaves the machine), and the Word add-in (shield a document's personal data with review from inside Word, restore later; desktop Word on Windows and macOS)
- Official YouTube channel: https://www.youtube.com/@Occlira
- Typical use cases: anonymize sensitive text before pasting it into ChatGPT, Claude or Gemini; redact documents before sharing them; blur faces and strip EXIF from photos; GDPR- and HIPAA-oriented local redaction for law firms, healthcare and therapists, accountants and consultants, journalists, and HR/compliance teams