How to redact a PDF, and why a black box does nothing

Covering is not removing — the five ways a black box fails

• Selection. Drag a cursor across the box and the text underneath highlights and copies. This requires no tools at all.
• Text extraction. Every PDF reader, every browser and a dozen command-line utilities can dump the text of a page regardless of what is painted over it.
• Search. The word you covered is still indexed. Searching the document for it takes the reader straight to the box.
• Copying the page. Copy the whole page and paste it elsewhere; the rectangle comes along, and so does the text behind it in most editors.
• Revision history. If the document was redacted by editing an earlier version and saving, the original text can still be present in the file as an unreachable object from the previous save.

A useful mental model: the rectangle is a new layer on top of the page. Nothing about adding a layer changes what is underneath it. Every one of the five failures above comes from the same fact.

How to check whether a document was really redacted

It takes about fifteen seconds, and you can do it in whatever reader you already have open. Do all four tests, because each one catches a different failure.

• Try to select the text under the box. If a selection appears, nothing was redacted.
• Search for a word you know is behind a box. If it is found, the content is still there.
• Select the entire page and copy it into a plain text editor. Redacted words that reappear were never removed.
• Look at the file size. A redaction that changed nothing except painting rectangles leaves the file the same size or slightly larger; genuine removal usually makes it smaller.

The third test is the revealing one. It bypasses the page rendering entirely and asks the file what text it contains, which is exactly what any determined reader would do.

What genuine redaction actually involves

A real redaction tool does the unglamorous work: it identifies the content in question at the level of the page content stream, deletes those drawing operations, removes the corresponding text from the encoding and the glyph resources, rewrites the file so no earlier revision remains, and only then paints whatever placeholder is wanted. The result is a document in which the removed words do not exist anywhere, including in the parts of the file you cannot see.

That is a different operation from drawing, and it needs a tool built for it. There is no shortcut, and there is no way to fake it by combining generic PDF utilities.

The four local methods, ranked by how well they hold up

If you do not have a redaction tool, some things you can do yourself are genuinely effective and others only look effective. In descending order of reliability:

• Delete the page. The strongest option available without special software, and the only one that leaves no trace at all. If the sensitive material is on a page of its own — an annex, a signature page, a schedule — deleting that page removes the content completely, because the page object itself is gone. Extract Pages does this locally.
• Rebuild the pages as images. Rendering each page to an image and rebuilding a PDF from those images destroys the text layer entirely. The words can no longer be selected, searched or extracted, because they are no longer text — they are pixels. Two honest caveats: the pixels are still there, so a low-resolution render is what makes small text unreadable rather than any guarantee; and you lose everything that made the file useful, including all text, links and form fields. PDF to JPG followed by JPG to PDF produces this.
• Cover the region at the content level before rendering. Draw an opaque rectangle over the text in a tool that writes it into the page content, then render the page to an image. This removes the reading path through a text layer, but it does not remove the pixels underneath in the rendered image if the rectangle fails to cover them exactly. It is a mitigation, not a redaction.
• Draw a black box in a PDF editor and save. This is the one that keeps making headlines. It changes nothing that matters.

If the document has legal weight or regulatory exposure, none of these substitutes for a real redaction tool, and none of them produces the audit record that a proper redaction workflow creates. Use them for everyday privacy housekeeping, not for disclosure obligations.

What people forget to redact

Even when the visible content is handled correctly, four other places in a PDF routinely carry the same information.

• Document metadata. The author, the original title, the creating application and the creation and modification timestamps stay with the file. A document whose author name was removed from the text can still name them in its properties. There is a guide on what a PDF says about you that covers this in detail.
• Comments and annotations. A comment can be collapsed, hidden or set to not print, while remaining in the file and fully visible to anyone who opens the comments panel.
• Attachments. PDFs can carry embedded files — a spreadsheet, a source document, a thumbnail. These survive page-level editing untouched, and a reader that lists attachments will show them.
• Hidden and optional layers. Some PDFs contain optional content groups that a reader can switch on. Content that is switched off is not content that is absent, and a preview of it may also exist as an embedded thumbnail.

A short checklist before you send anything

• Decide whether this is housekeeping or a disclosure obligation. If it is the second, use a dedicated redaction tool and keep its audit record.
• Remove rather than cover. Delete the page if you can; rebuild the pages as images if you cannot.
• Check the document properties, the comments panel and the attachment list, not just the page.
• Re-open the saved file in a different viewer and run the four checks again — select, search, copy, compare size.
• Keep the original. Every one of the methods above is destructive, and you will want the unredacted source for your own records.