How redactions fail
A PDF page is built in layers of instructions: draw this text here, draw this image there. When someone "redacts" by drawing a black rectangle, the rectangle is just another instruction drawn after the text. The text is still in the file. Anyone can select the area and copy the hidden words, search the document and find them, or open the file in an editor and delete the rectangle. The same applies to highlighting text in black, setting the text colour to white, or placing an image over it. Real documents released by courts, governments and companies have been unredacted this way.
Other hidden places also leak: document metadata (author and title), comments and annotations, earlier versions kept by incremental saves, invisible OCR text layers behind scanned images, form field values, and bookmarks with descriptive titles.
What proper redaction does
Proper redaction removes the content and anything that could reconstruct it. Professional tools do it by editing the page's content to delete the text and image data under the box, then cleaning up metadata and hidden data. That is effective but complex; mistakes happen with unusual fonts, text split across several instructions, or images that partly overlap the box.
The redaction tool on this site takes a simpler, more robust approach. Every page with a box is rendered to an image, the box is painted onto the image, and the page is replaced by it. The page's original text, images and hidden layers are not carried into the new file at all, so there is nothing to recover. The trade-off is that the rest of that page's text is no longer selectable. Pages you did not redact stay as they were, unless you choose to rasterise the whole document, which strips hidden content from every page. Document metadata is removed either way.
A redaction checklist
- Names, including in signatures, letterheads, email headers and footers.
- Addresses, phone numbers and email addresses.
- Dates of birth, and identity, tax, Medicare, account and policy numbers.
- Faces and identifying features in photos.
- File references and internal codes that could identify a person.
- Handwritten notes and stamps in margins.
- Every page, including annexures and attachments merged in.
Verifying a redaction
- Select text on each redacted page. With this tool there will be none: the page is an image.
- Search the file for each redacted term, such as a surname or a number. It should not be found.
- Check the properties (File → Properties): title and author should be empty.
- Zoom in on the edges of each box for characters peeking out.
PDF Pro automates the first two checks: it re-reads the output, confirms that redacted pages contain no extractable text, searches the whole file for the terms you list, and gives you a report to keep with the document.
Before releasing
Keep an unredacted original in a secure place; redaction cannot be reversed. Have someone else check the redacted copy against the checklist if it matters. If you are responding to a Freedom of Information request, a subpoena or a privacy request, follow the specific rules that apply; some require you to mark the grounds for each redaction, which you can add as text with the Sign PDF text tool before redacting. Because this tool runs in your browser, the unredacted document never leaves your device during the process.
Redacting scanned documents
Scanned documents are images, so a box drawn on top seems safe. It often is not, because many scans include an invisible OCR text layer that makes them searchable, and that layer still holds the words under the box. Rasterising the page, as this tool does, removes the OCR layer along with everything else.