How to redact a PDF before sharing

The difference between covering and removing

A black box drawn on a page is an annotation. It sits on top of the text and hides it from the eye, but the text underneath is still in the file. Select the area, copy, paste into a text editor, and the "redacted" content appears. Tools that strip annotations — or a recipient who simply moves the box — recover the original. That is not redaction; it is decoration.

Real redaction removes the content from the file's byte stream. The characters, the image regions, and the metadata that referenced them are gone, not just hidden. The result is a file where the redacted information does not exist anymore, and no selection, no annotation removal, and no forensic tool can bring it back.

Three things you must redact, not just one

Redaction is not only about visible text. A PDF can leak information in three layers, and a thorough redaction addresses all of them.

• Visible text and images: the words and pictures on the page, including text under cover boxes.
• Metadata: author names, creation dates, editing history, comments, and sometimes earlier redactions that were not burned in.
• Hidden content: embedded files, attachments, hidden layers, form fields, and the document's structure tree.

The safe workflow: remove, do not cover

If you do not have a dedicated burn-in redaction tool, the safest path is to remove the content you must hide at the source, then produce a clean PDF. The practical steps below work with the tools in a typical browser PDF toolkit.

• Identify exactly what must not leave the document. Names, account numbers, dates of birth, addresses, signatures.
• If the sensitive content is on whole pages, remove those pages and keep only what the recipient needs.
• If the sensitive content is a small region, crop the page to exclude it, or replace the source with a version where the region is blanked before you export to PDF.
• Run a metadata-removal pass so author names, comments, and edit history do not travel with the file.
• Flatten any form fields so their values and labels are baked into the page rather than live fields.
• Save the result as a new file; never overwrite the original you may need later.

Removing pages is the cleanest redaction

When the sensitive content occupies a whole page or a whole section, removing those pages is the most reliable redaction. There is no text to recover, no box to move, no metadata to strip from the region. The pages simply are not in the file.

This is the right approach for a document that contains a cover sheet with personal details, a signature page that should not travel with the body, or an appendix that the recipient does not need. Use a remove-pages tool, verify the new page count, and confirm the sensitive page is gone by jumping to the page numbers around it.

Cropping and region removal

When the sensitive content is a region on an otherwise-needed page, cropping is the practical option. Crop the page to exclude the region, or crop tightly around the content you want to keep. The cropped-out area is not rendered, and the result is a page that does not contain the redacted region.

Cropping is stronger than a cover box because the region is not in the rendered output, but it is weaker than true burn-in redaction on the byte stream — a determined forensic analysis of the file may recover the original page objects in some toolchains. For most sharing scenarios, crop-plus-metadata-removal is more than enough. For legal discovery or regulated material, use a dedicated burn-in redaction tool that removes the content from the byte stream and verifies the result.

Metadata and comments leak more than you think

A document that looks clean on the page can carry author names, editor identities, creation and modification dates, comments, and tracked changes in its metadata. Before sharing, run a metadata-removal pass. Open the result in a reader that exposes metadata and comments, and confirm nothing sensitive remains.

A common failure is redacting visible text but leaving the author name in the document properties, or leaving an earlier redaction as a removable annotation rather than a burned-in removal. The metadata-removal step catches the first; the remove-don't-cover principle catches the second.

How to verify a redaction before you send

Trust the verification, not the redaction. Before you share the file, run these checks.

• Select the area that was redacted and copy it. Paste into a text editor. If the redacted text appears, the redaction failed.
• Open the document properties and read the metadata. If author names or comments remain, run the metadata-removal pass again.
• Open any comments or annotations panel. If cover boxes are listed as removable annotations, they are not burned in.
• Check the page count against the source. If pages were removed, the count should reflect it.
• Open the file in a second reader, not the one you used to redact. Different readers expose different metadata and hidden content.

Scanned documents: redact the image, not the text

A scanned PDF that has been OCR'd has both an image and a text layer. Redacting only the text layer leaves the image intact — a recipient can still see the scanned name. Redacting only the image (with a black box) leaves the text layer intact — a recipient can still select and copy the name from the OCR text.

For a scanned document, you must redact both layers. Remove or crop the image region, and remove the corresponding text from the OCR layer. The cleanest path is to redact the source before OCR, or to crop the page so the sensitive region is not in the rendered output, then re-run OCR on the cropped page so the text layer does not contain the redacted words.

When to use a dedicated redaction tool

For legal discovery, regulated material, and anything that could end up in front of a court or a regulator, use a dedicated redaction tool that burns the redaction into the byte stream and produces a verification report. Those tools guarantee that the removed content is gone, not just hidden, and they document the redaction for the record.

For everyday sharing — a contract with a personal detail removed, a receipt with a card number cropped, a memo with a name taken off — the browser-toolkit workflow (remove pages, crop, strip metadata, flatten, verify) is sufficient and keeps the file on your device. Match the tool to the stakes.

A redaction checklist before you share

Before you send a redacted PDF:

• You removed or cropped the sensitive content; you did not just cover it.
• You ran a metadata-removal pass and confirmed author names and comments are gone.
• You flattened form fields so values are baked in, not live.
• You selected the redacted area and copied it; the redacted text did not appear.
• You opened the file in a second reader and checked the document properties.
• For a scan, you redacted both the image and the OCR text layer.