How to anonymise a PDF step by step (2026)

If you are looking for how to anonymise a PDF, you probably need to publish, send or archive a document without personal data in it. How to do it well depends on volume: a single document can be reviewed carefully by hand; a weekly notice, a case file or a recurring bulletin needs a process that does not rely on nobody forgetting a field.

Why is blacking out text not anonymisation?

The most common mistake — and the one behind most fines, as covered in our review of real AEPD cases — is covering the data with a coloured shape in the PDF editor or in Word before exporting. In most tools that shape sits on top of the text, not instead of it: the original content still exists in the text layer and can be recovered by copying it, extracting it with a simple script or even moving the object that covers it.

Anonymising a PDF properly means removing the data from the text layer, not hiding it visually. For the full conceptual framework, see what document anonymisation is.

How do you anonymise a PDF manually, step by step?

1. Identify every personal data point, not just the name: ID number, address, phone, IBAN, licence plate, signature, date of birth and anything that combined can identify someone.
2. Use the editor real redaction tool (in Adobe Acrobat Pro it is called "Redact", not "Highlight" or "Draw shape"), which removes the underlying text as well as covering it.
3. Apply the redaction and save a new copy: in most programs the process is not final until changes are "applied" and the file exported.
4. Review file metadata (author, company, comments, revision history, GPS coordinates in images) and remove it. See how to remove PDF metadata.
5. Verify the result by trying to select and copy the text in redacted areas. If any of the original data can be copied, the process failed.
6. Repeat the check after any format conversion (Word to PDF, OCR scanning, merging PDFs). Each conversion can reintroduce data you had already removed.

When does the manual process stop being viable?

• Inconsistency between people: whoever reviews the document on Monday does not apply the same criteria as on Friday.
• Mixed formats: PDF, Word, Excel, CSV, scanned images — each hides data in a different place.
• No audit trail: nobody can later prove what was detected and removed in each document.

What to look for if you automate it?

• It must remove data from the text layer, not cover it visually.
• Format coverage: PDF, Word, Excel, CSV, JSON and scanned images with OCR, without manual exporting and reprocessing.
• Where processing happens: for the public sector, in a controlled and auditable environment, relevant to the National Security Framework — see anonymisation in public administration.
• A record of what was detected and removed, to justify the processing in an audit or inspection.
• Consistency: the same criteria applied to one hundred documents as to one.

Anonimizia targets exactly that scenario: automatic detection across documents and batches, human review before export and a record of the process. You can request a trial.

Frequently asked questions

Is exporting Word to PDF after deleting the data enough?

Only if you actually deleted the text (rather than covering it) and also cleaned the metadata and tracked changes in the original Word file, which may still contain the data.

Can OCR reintroduce data I had already removed?

Yes. If you scan a document where the data was only visually covered but the text remained underneath, OCR can capture it again as readable text. Always check the final output.

Does anonymising a PDF affect its electronic signature?

It can: modifying an electronically signed document invalidates the signature. Anonymise before signing, or produce and sign an anonymised version as a new document.

Sources and references

• AEPD — Anonymisation and pseudonymisation
• Real AEPD fines for failing to anonymise documents