Why effective PDF redaction requires the right tool
A PDF can look redacted even though the covered text remains in the file. This can happen when drawing or annotation tools are used to hide data. The recipient may then be able to recover the information, for example by copying the text.
The UK Information Commissioner’s Office (ICO) describes this problem in its guidance on safely disclosing information. The guidance explains why visually covering text and saving a document as a PDF do not, by themselves, ensure that the content has been permanently removed.
When choosing a solution, establish whether it removes the content selected for redaction and produces an output file for sharing. The appearance of a black box does not establish whether this process has been effective.
What a Polish Supreme Administrative Court ruling means for document security
In a statement published on 26 June 2026, Poland’s Personal Data Protection Office (UODO) discussed the position of the Supreme Administrative Court on regularly testing, assessing and evaluating the effectiveness of security measures. The judgment of 7 May 2026, case reference III OSK 2694/23, concerned the protection of a telecommunications operator’s subscriber data.
The case does not set out a procedure for checking PDF files. However, it offers a practical lesson for organisations that share documents: they need to understand how their chosen safeguards work and assess their effectiveness. For redaction, this includes choosing a solution that permanently removes the selected content.
How to check PDF redaction when reviewing your workflow
If your existing process relies on manually covering data in an editor, you can begin reviewing it with three simple checks:
- Open the saved copy. Check the file intended for sharing, rather than the preview of selected areas in the editor.
- Search for and copy the covered text. If you can recover it, the information remains accessible. In a scan without a text layer, a search that returns no results is inconclusive.
- Check for overlays. If moving or removing a box in a working copy reveals the data, the method needs to change. Being unable to edit the box does not, on its own, prove that the content has been permanently removed.
These checks can reveal common problems with visually covering information. They do not replace a review of the entire document, its hidden information or its context. Use the findings to improve your process and choose an appropriate tool.
What the user still needs to decide
Before approving a document, consider what information its recipient may receive. A name, address and signature may be only some of the data that needs to be removed. A description of circumstances that could identify someone may also matter.
UODO highlighted this issue in its statement on anonymising handwritten documents. In the material published on 7 July 2025, it noted that handwriting may, in certain circumstances, make it possible to identify the author even when other data has been covered.
In practice, first establish whether the document will be shared with a specific recipient or published online. Then approve the appropriate selection of areas and consider whether the remaining details are necessary for that purpose. Information outside the selected areas, including the filename, metadata and attachments, also needs separate consideration.
Are empty search results enough to confirm effective redaction?
No. The document may be a scan without a text layer, and identifying information may also exist outside the text being searched. When choosing a tool, consider how it permanently removes data. When preparing a copy, make sure the scope of redaction is appropriate too.
How Bluur permanently removes content from a document
Bluur handles the technical removal of content from approved areas of a document. AI suggests areas belonging to defined data categories. The user reviews the selections, adds any further areas and approves the redaction. The system then permanently removes the content of the selected areas and prepares the document for download.
In the output file, a blacked-out area is not a movable cover with the original content preserved underneath. This means users do not need to carry out the technical removal manually or treat every black box as an overlay requiring a separate test.
You decide what needs to be removed. Bluur permanently removes the content of the approved areas. This division simplifies everyday document work and allows users to focus their review on the appropriate scope of redaction.
An example of preparing a document in Bluur
Suppose an organisation wants to share a form containing a name, email address and signature. This is an illustrative workflow, rather than the result of testing a specific file.
The user uploads the document to Bluur, reviews the classifier’s suggestions and approves the areas whose content should be removed. If additional areas need redaction, the user adds them. The system applies the redaction and makes the output available for download.
The user’s work centres on selecting the information to remove. Bluur’s mechanism handles the permanent removal of that content from the prepared copy. Before sending it, the user still needs to ensure that the correct version is being shared and that the scope matches the purpose of disclosure.
How to streamline document redaction across your organisation
A repeatable process starts with defining which data to remove for a particular use. Publishing a document online may require broader redaction than providing a copy to an authorised recipient. Establish these rules before starting work on the files.
Bluur combines classification suggestions with permanent removal of content from approved areas. This allows the team to focus on deciding what may be shared, rather than manually creating and checking visual covers.
Try Bluur and explore the workflow using a test document with fictional data. Choose material similar to the documents you work with: a form, a contract or a scan containing signatures and handwritten notes.
Do you need to check every black box after redacting a document in Bluur?
Bluur permanently removes the content of approved areas. You do not need to treat these areas as manually drawn covers and separately check whether they can be moved. The user still reviews the selected areas and decides what information is appropriate for the purpose of sharing.
Does AI decide on its own what to remove?
AI suggests areas belonging to data categories. The user reviews the suggestions, adds to them where needed and approves the scope of redaction.
