Format guide · .pdf
Anonymize a PDF in your browser
Anonymize any document before it reaches an AI. Nothing leaves your device. Text is extracted from the PDF, anonymized, and returned as an editable Word file plus Markdown; the original PDF is never modified.
- 0 network requests
- Works with Wi‑Fi off
- Free, no account
- Same format back
The tool below is preset for .pdf files. Up to 10 files, 50 MB each.
- 1Upload
- 2Choose rule
- 3Review
- 4Download
Upload
Drop your PDF file here or choose
Up to 10 files · 50 MB each · files stay on your device
Old .doc, .ppt and .xls files: save them as .docx, .pptx or .xlsx first.
What stays intact
What we preserve in PDF files
The file is edited in place: only the detected text changes. Everything else is written back byte for byte where possible.
Text content and reading order
Text is extracted page by page with pdf.js in your browser and rebuilt into paragraphs.
Headings and bullets
Font size and position are used to reconstruct headings and bulleted lists in the DOCX and Markdown output.
Output as DOCX + Markdown
You get an editable Word file and a Markdown version that is ready to paste into a chat.
Page breaks
Each page starts a new section in the DOCX so long documents stay navigable.
Not preserved: exact layout
Multi-column layouts, forms and images are not reproduced. This is a text-first conversion.
Not supported: scanned PDFs
Image-only PDFs contain no text layer. Run OCR first, then anonymize the resulting DOCX or text.
What gets detected
Names, numbers and identifiers: three layers
Patterns with checksums (IBAN, national IDs, card numbers, emails, phones, URLs, IPs, dates, plates), your own dictionary of names and terms, and an optional model that runs in your browser for person, company and place names.
- Person names
- Companies and brands
- Places and addresses
- Emails and phone numbers
- National IDs, tax IDs, passports
- IBANs and card numbers
- URLs and IP addresses
- Dates and license plates
Questions
PDF questions, answered
Why do I get a DOCX instead of a PDF back?
Editing text inside a PDF reliably, without leaving the original under a black box, is not something a browser can do for arbitrary files. Extracting the text and returning an editable DOCX plus Markdown is safer and more useful for pasting into an AI.
Are scanned PDFs supported?
No. A scanned PDF is an image without a text layer. Run OCR in another tool first, then anonymize the resulting Word or text file here.
Will the layout look like the original?
Headings, paragraphs and bullet lists are reconstructed. Columns, forms and images are not reproduced.
Is the PDF uploaded to a server?
No. Extraction uses pdf.js inside your browser and the anonymization runs on your device. There are 0 network requests during processing.
Can it handle password-protected PDFs?
Not currently. Remove the password in your PDF viewer first, then drop the file in.
How large can the PDF be?
Up to 50 MB per file, up to 10 files per run. Very long PDFs simply take longer to extract.
Other formats
Same tool, eight more file types
Drop any of these into the tool above; the format is detected from the file, so you never have to switch pages.