How to redact a PDF safely, and prove the text is really gone

The short answer: a black box is not redaction

A PDF page is built from separate layers of instructions: text objects, images, vector shapes and annotations. When you add a black rectangle with a highlighter, a shape tool or a comment tool, you add one more object on top of the page. The text object underneath is untouched. Anyone can open the file, press Ctrl+A and Ctrl+C, paste into a text editor, and read what you meant to hide. Some people simply drag the box aside or delete it in an editor.

Real redaction means the underlying content is removed or replaced, not covered. There are two honest ways to achieve that. One is a redaction feature in a professional editor that rewrites the content stream and deletes the text inside the marked area. The other is to flatten the page into a picture so that there is no text object left to extract at all, and then draw the box. The second method is blunt, but it is easy to verify, which matters when a mistake could expose an ID number or a medical detail.

If you can still select or search the covered text, the redaction failed.

Secure redaction versus visual redaction

It helps to name the two modes clearly, because the difference is the whole point of this article.

The trade-off of the secure method is that the whole page becomes an image. You lose selectable text everywhere, including the parts you did not redact, and links and form fields stop working. For a one-off document such as a bank statement or a signed contract that you are sending out, that is usually acceptable. For a long report that people need to search, it may not be, and you would be better off using a desktop editor with a true redaction feature on the original file.

How to redact a PDF with the Redact PDF tool

The Redact PDF tool on this site runs entirely in your browser, so the file is not uploaded anywhere. Its default mode is the secure one described above. The steps are:

1. Open the Redact PDF tool and choose your file. Thumbnails or a page view appear.
2. Drag on the page to draw a box over each area to hide: names, account numbers, addresses, signatures, faces in scanned photos. You can move, resize or delete a box. On a phone, switch to scroll mode to move the page, then back to draw mode.
3. Choose Secure mode (the default) and set the resolution and JPEG quality. 150 dpi is a reasonable starting point for reading on screen; 200 dpi or more is better if the text is small.
4. Press Apply and wait while each page is rendered.
5. Read the text-extraction check the tool shows. It attempts to pull text out of the result; for a secure redaction the page should return no text at all.
6. Download the redacted PDF, then run your own checks as described below before you send it.

Redact the copy, never the original, and keep the original somewhere private.

How to verify the text is really gone

Never trust the look of the page. Verification takes a minute and is the step most people skip. Try several independent checks on the downloaded file, not on the one you were editing.

If any of those checks still shows text, discard the file and start again from the original with larger boxes or secure mode.

• Select all (Ctrl+A or Cmd+A) in your PDF viewer. In a flattened file nothing highlights as text; the page behaves like a photograph.
• Use the viewer search (Ctrl+F) for a word you redacted. It should find zero matches.
• Run the PDF to Text tool on the redacted file. The output should be empty or contain only text you intended to leave, if you used a method that keeps some text.
• Open the file in a second viewer or a browser, in case the first one hides something.
• Zoom in to 400 percent and look at the edges of each box. A box that is a few pixels too small can leave the top of a letter visible.

Common leaks that survive a redaction

Even when the visible page is clean, other parts of a PDF can carry the same information. These are the ones that catch people out.

• Metadata. The document information fields such as Title, Author, Subject and Keywords often contain a person or company name. A title like "Settlement offer - John Smith" defeats the purpose of hiding the name inside the page.
• Hidden layers and annotations. Comments, sticky notes, form-field values and optional content layers are stored apart from page text. Flattening to an image removes them from the page, but check any copy you produce by another route.
• Bookmarks and outlines. A table of contents entry can repeat a redacted heading word for word.
• Attachments. A PDF can contain embedded files that are not shown on the page.
• File name. "contract-redacted-for-Mr-Patel.pdf" is itself a disclosure.
• Previous versions. Some editors save incremental updates, so an earlier version of the page can still sit inside the file. Rebuilding the pages as images avoids this.
• Images with their own text. A scanned page is already a picture; redaction boxes work on it, but remember that a screenshot elsewhere in the document can contain the same data.

Redaction covers the page; metadata, bookmarks and attachments need separate attention.

Clean the metadata after redacting

After you redact, open the result in the PDF Metadata Editor on this site. It shows the title, author, subject, keywords, creator, producer and dates, and has a button that removes all of them in one click. Because the secure mode already creates a fresh document from page images, the original author fields usually do not carry over, but you should still look rather than assume. Read the fields, clear anything that identifies a person, and save.

If you also need to prevent anyone from adding or changing things later, remember that a PDF password does not protect against copying a flattened image. For a final deliverable, it is normal to combine the steps: redact in secure mode, clean the metadata, rename the file neutrally, and check by trying to extract text one last time.

When this method is not enough

Be honest about the limits. Flattening to an image does not make the picture itself anonymous: a handwriting style, a distinctive layout or a barcode that you forgot to cover still identifies a document. It also does not apply to information elsewhere, such as a covering email. If you work with legally sensitive disclosure, litigation or regulated data, use the redaction workflow your organisation requires and have someone review the output. For long documents that must stay searchable, use a desktop editor that performs true content removal and offers an audit trail, and verify that output with the same checks above.

Frequently asked questions

Try these tools