Personal data
The identity of people named in the document.
Anonymize a document: PDF, Word or scan
Make any document confidential in seconds.
- No document is ever stored
- Layout preserved
- Exported flattened
Found and masked automatically
NamesEmailsPhonesAddressesOfficial IDsDatesIBAN / RIBBICCard numbersTax IDsAmounts with a currencyPasswordsOne-time codesSSH keysAWSStripeOpenAI / AnthropicGitHubJWTSIRENSIRETVAT numbersInternal URLsIntranet linksPrivate hostnames
ONYRI Sanitize anonymizes a PDF, Word or scanned document in your browser. It finds names , ID numbers and bank details , then gives you back a masked copy with the layout intact. The file never leaves your device.
How to anonymize a document in three steps
Drop
Drag a PDF, Word file or scan into the browser. Scans are read by on-device OCR.
Review
Every detected item is shown on the page. Untick a false positive, select a missed word or draw an area over a signature before you export.
Download
Download an anonymized copy as a PDF, or as a PNG for an image. The masks are flattened into the file, so the text is really gone.
About 40 detectors across 6 families.
Detection is country-aware (FR, US, GB, BE, DE/AT/CH). You can add your own rules and profiles from the app.
Financial
Bank and money details.
Credentials
Anything that opens a door.
Technical secrets
API keys and tokens left in exports and logs.
Corporate
Company identifiers.
Strategic
Links that reveal your internal systems.
Built for the documents you actually handle.
Each page explains what is detected, what is masked and what to check before you share.
Invoices & financial documents
Bank details, amounts and tax IDs covered.
Read the guide : Invoices & financial documentsAnd for every format:
Your file never leaves your device.
Detection, masking and export all run inside your browser. Our servers only see your account and a document count. Never the content.
- detectors, including names, IBANs and ID numbers
- ≈40
- data families
- 6
- countries recognized, from Social Security numbers to VAT IDs
- 11
- formats: PDF, Word and scans
- 3
The file stays on your device
It is opened in the browser's memory and nowhere else.
Nothing is uploaded
No server receives the document, the text or the masks.
Token mapping lives in memory only
In token mode, the value ↔ token table stays in the tab and vanishes when you close it.
Verify it yourself
Open the Network tab of your browser's DevTools while you anonymize. You will see a usage count being sent. Never the document or its text.
Which method should you use to anonymize a document?
There are four common ways to hide the data in a document. They do not protect it in the same way.
| Method | What happens | Good to know |
|---|---|---|
| A black box in Word or in a PDF | The box hides the text on screen, but the text stays in the file. | Text recoverableA simple copy and paste brings it back. |
| PDF software with a redaction feature | The text is removed where you mark it. | Needs reviewEverything is manual: a missed name stays readable. |
| An online service that receives the file | The document is sent to a third party's servers to be processed. | Sent to a third partyYou hand the document to that service, under its own terms. |
| ONYRI Sanitize | Detection, masking and the flattened export happen in your browser. You finish by hand. | Needs reviewDetection is not exhaustive: review the preview before you download. |
Start today. Upgrade when you need more.
Every plan runs 100% in the browser. Pro adds token mode, technical-secret detectors and unlimited documents.
MonthlyAnnual
Free
To try ONYRI on a few documents.
$0
- 3 documents per day
- 5 pages per document
- Black marker mode (irreversible)
- Base detectors: names, emails, phones, IBAN, IDs
- 3 custom rules
- 100% in your browser — no file is uploaded
Pro
For professionals who handle sensitive documents every day.
$11.90
- Everything in Free
- Unlimited documents and pages
- Token mode: readable pseudonymization
- Technical secrets: API keys, tokens, SSH keys
- Unlimited custom rules and profiles
- Standard email support
Team
The Pro engine on every seat. Team features are rolling out: contact us before you subscribe.
$28.90
- Everything in Pro, for each seat
- Billed per user, 3 seats minimum
- Priority email support
- DPA available
- Coming soon: shared workspace, invitations and roles
- Coming soon: audit logs
Lifetime
Pay once and keep Pro or Team. No subscription, no renewal.
$229
- Everything in Pro, paid once
- Team lifetime: everything in Team, 3 seats
- Future updates to your plan included
- No charges after that
- VAT invoice, card payment
Learn to make a document confidential
Methods, mistakes to avoid and what the law expects. Written to be useful even if you never use our tool.
Why blacking out text in a PDF often fails
A black rectangle covers text but does not remove it. Here is how hidden words come back, what real redaction does, and how to test your own file.
Read articleHow to redact a PDF properly: a step-by-step guide
Covering a name with a black shape leaves the text in the file. Here are the routes that really remove it, and a short routine to test your PDF before you send it.
Read articleAnonymization, pseudonymization, redaction: what is the difference?
Anonymous data leaves the GDPR. Pseudonymized data stays in it, at least for whoever holds the key. Redaction is what you do to the page. Here is how to tell them apart.
Read article
Straight answers.
Can't find your answer?
Email usDoes my file leave my computer?
No. The document is opened in your browser. Detection, masking and export run on your device. Our servers only receive your account details and counters: how many documents and pages, which types of data were masked, the country and language recognised. They never see the content. You can check it in the Network tab of your browser's DevTools.
Is this GDPR anonymization?
Be careful with the word. Black-marker mode removes the text from the exported file for good. Token mode replaces values with consistent labels such as [NAME1]: that is pseudonymization, not anonymization in the legal sense, because anyone who holds the original can link the labels back. Detection is based on patterns and is not exhaustive, so review the preview before you download. ONYRI Sanitize reduces your exposure. It does not guarantee GDPR compliance on its own.
Which formats and limits?
PDF, Word (.docx) and images (PNG, JPG, WebP). Scans, photos and scanned PDF pages with no text layer are read by OCR on your device, in English, French and German, and also in your browser's language among some thirty others (Spanish, Portuguese, Polish, Arabic, Japanese…). Detection recognises ID numbers, phone numbers, addresses and dates from 47 countries. The interface is in English, French and German: your browser can translate it. A Word file is laid out as A4 pages and exported as a flattened PDF. The Free plan allows 3 documents per day (UTC reset) and 5 pages per document. Pro removes those limits. Scans take longer because OCR runs on your machine, page by page.
Can the masking be undone?
Not in the exported file. The export is flattened in both modes: the text is removed, not hidden under a layer. Black marker leaves a solid bar. Token mode (Pro) leaves a label such as [NAME1], and the same value always gets the same label. Only someone who holds the original document can link a label back to a value. ONYRI never stores your file, so keep your original.
Does it work on scanned documents?
Yes. A scan, a photo or a scanned PDF page with no text layer is read by OCR that runs inside the browser. The text found is then checked by the same detectors. OCR accuracy depends on the scan quality, and handwriting is not read. Check the preview for anything that was missed.
Can I add my own rules?
Yes. Custom rules (keywords or patterns) and profiles let you mask project names, client codes or anything specific to your work. The Free plan includes 3 custom rules. Pro and Team make them unlimited. Sharing rules across a team workspace is coming soon.
Is there a plan without a subscription?
Yes. The Free plan lets you anonymize 3 documents a day, up to 5 pages each, in Black marker mode, with manual masking. Pro removes these limits and adds Token mode and technical-secret detection. Prices are on the Pricing page.
What if the detection misses something?
You mask it by hand before you download. Select a word to mask every occurrence, search for a term across the document, or draw an area over a signature, a stamp or a photo. On a computer, a name-recognition model also runs on your device to find the names of people and companies. It does not run on phones or tablets. In languages without a names model (Polish, Czech, Russian, Japanese, Korean…), a name is only masked when it follows a title or a label, or sits in a signature: review the preview and mask anything left by hand.
Ready to make a document confidential?
Nothing to install. Open your browser, drop the file, download it anonymized.