Ways to anonymize a PDF: what actually works
Anonymizing a PDF for free is perfectly possible. The question is whether the technique you use is genuinely irreversible, and whether it leaves an audit log good enough to prove compliance to a supervisory authority. These are the most common routes, with their real limits.
The six most common routes
Free Adobe Acrobat Reader, with manual editing
Open the PDF, draw black rectangles over the personal data, save.
- It is already installed on almost every machine
- It costs nothing
- The free version has no redaction tool: those rectangles are graphic annotations drawn on top of the PDF
- The original text is still underneath and comes back by selecting, copying and pasting into another document
Unacceptable as anonymization. It is the number one cause of data leaks in documents believed to be clean.
Adobe Acrobat Pro
The "official" tool in many organizations. It is not free —around €23 a month— but it is included here because it is what most people already have.
- It includes real redaction, which removes the underlying text
- It has a metadata cleaning tool
- Redaction and metadata cleaning are two separate steps, and skipping one is common
- It does not detect what should be redacted: the user marks every occurrence by hand
- It does not recognize Spanish identifiers —DNI, NIE, social security numbers— as automatic patterns
- It produces no exportable audit log to show a supervisory authority
Technically this is real anonymization, but the process is slow and prone to human error: it is easy to miss occurrences or forget the metadata.
LibreOffice Draw
Open the PDF in Draw, delete the text objects holding personal data, export back to PDF.
- Free, with no monthly document limit
- It genuinely deletes the text rather than covering it
- It only works on editable PDFs, not on scans
- There is no redaction mode: text is deleted by hand, one item at a time
- It does not clean the exported PDF metadata, which stays in the XMP
- Unworkable for batches of documents
- Changes are not recorded for audit purposes
Useful for one-off cases, such as a single contract. Unworkable for a repeating process.
Free online tools such as PDF24, Smallpdf or iLovePDF
Upload the PDF, apply the online redaction tool, download the result.
- No installation required
- Some of them do redact the underlying text correctly
- Files are uploaded to third-party servers: if they hold personal data, the upload is itself processing, requiring a data processing agreement and probably an impact assessment
- Most have no EU servers and sign no GDPR-compliant processing agreement
- Size and daily document limits in the free tier
- They produce no exportable audit log
Legally problematic for documents holding real personal data. If used at all, it should be on already-anonymized documents, never on the originals.
In-house Python scripts with PyPDF2, pdfplumber or pikepdf
Write a script that opens the PDF, detects personal data patterns by regular expression or NLP, replaces them and saves a clean PDF.
- Full control over the process
- Reproducible and auditable if designed well
- Free if a developer is already on staff
- It requires programming skills and development time
- Building an identifier detector accurate enough to tell a national ID number from any other number in the text is not trivial
- PDFs vary enormously in structure: a script that works on one type fails on another
- It needs continuous maintenance whenever document formats change
Reasonable only where a technical team already exists and the volume justifies the build. For most companies, maintaining an in-house script costs more than paying for a specialized tool.
pdf-redact-tools, from the Freedom of the Press Foundation
Open-source tool aimed at journalism. It applies real redaction and converts pages into images to strip hidden layers.
- Free software, with a maximum-security approach
- Real redaction and removal of hidden layers
- Works on scanned documents
- It does not detect what to redact: it has to be paired with a separate detection tool
- Converting to images costs text selection, and with it accessibility
- It requires the command line
- Little recent development activity
Excellent for maximum-security redaction, the kind a journalistic leak demands. Overkill for an ordinary business workflow.
Summary table
| Tool | Real redaction | Removes metadata | Automatic detection | Works on scans | Audit log | GDPR frictionless |
|---|---|---|---|---|---|---|
| Free Adobe Reader | No | No | No | No | No | No |
| Adobe Acrobat Pro | Yes | Manual | No | Yes, with OCR | No | Partial |
| LibreOffice Draw | Manual | No | No | No | No | Partial |
| Online (PDF24, Smallpdf…) | Depends | No | No | Depends | No | No, outside the EU |
| Python script | Yes if implemented | Yes if implemented | Yes if trained | Yes, with OCR | Yes if implemented | Yes, in-house |
| pdf-redact-tools | Yes | Yes | No | Yes | No | Yes, in-house |
| anonimiza.do | Yes | Yes | Yes, EU-trained AI | Yes, OCR built in | Yes | Yes, EU and DPA |
Content carried over from our comparison published in April 2026. None of these cells replaces a test with your own documents.
When a professional tool pays off
- When the volume is recurring rather than one document a month
- When compliance has to be demonstrated and an exportable audit log is required
- When documents arrive scanned and need OCR
- When the identifiers are Spanish —DNI, NIE, IBAN, licence plates— and no generic tool recognizes them as a pattern
- When uploading the original to a third-party server outside the EU is not an option
Frequently asked questions
Can you anonymize a PDF for free?
Yes, perfectly. LibreOffice Draw genuinely deletes text in editable PDFs, and pdf-redact-tools applies real redaction to scans. What no free route offers is automatic detection of Spanish identifiers or an exportable audit log, which is what a supervisory authority asks for as proof of compliance.
Why is covering data with a black rectangle not enough?
Because in the free version of Adobe Reader that rectangle is a graphic annotation placed on top of the PDF, not a redaction. The original text is still underneath and comes back by selecting the whole document, copying and pasting into another file. It is the number one cause of data leaks in documents believed to be clean.
Is there a risk in using free online services with real data?
Yes. Uploading a document holding personal data to a third-party server is processing in itself: it requires a data processing agreement and probably an impact assessment. Most of these tools have no EU servers and sign no GDPR-compliant agreement. If used at all, it should be on already-anonymized documents.
Is it worth writing an in-house Python script?
Only where a technical team already exists and the volume justifies the build. The hard part is not opening the PDF; it is building an identifier detector accurate enough to tell a national ID number from any other number, and maintaining it as document formats change. For most companies the script costs more to maintain than the tool costs to buy.
Fuller analysis, route by route: Free PDF anonymization tools: an honest 2026 comparison