Redacting a PDF: covering text is not the same as removing it

A black box can leave the words underneath

Drawing a black rectangle over a number in a PDF editor looks convincing on screen. The characters beneath it are still part of the file, though, so anyone who selects that spot and copies it, or searches the document, may see them again. Several well-known document leaks happened in just this way.

The real question is therefore not whether the area looks black but whether the text data is gone from the file. The method below rebuilds each marked page as a flat picture, so there is nothing left under the box to select or copy.

What can be found automatically

Under ‘Find automatically’ you keep ticked the kinds of data to search for: ‘Email’, ‘Phone number’, ‘Card number’ and ‘US SSN’. Each is matched by its usual shape, so a number written with or without dashes is still caught in most layouts.

Anything without a fixed shape, such as a name, a street or a school, goes into ‘Words to find’, separated by commas. Every occurrence is marked at once. Treat the automatic result as a set of suggestions: a human check is a required step, not an optional extra.

When you enter names or addresses as words, also list the spelling variants you can think of, such as the spaced and unspaced forms, to reduce misses.

Step by step, from search to saved copy

Choose the PDF or drop it onto the page, tick the kinds you want, and press ‘Find and mark’. A box appears on every match and a line above the pages counts the areas and pages that will be redacted. Scroll through each page, drag across anything the search missed to draw a box, and press × on a box you do not want.

If you would rather begin again, ‘Clear all marks’ resets everything. Leave ‘Also remove document info such as author and title’ ticked if the file’s hidden properties should go too, then press ‘Redact and save’ followed by ‘Download’. The redacted copy keeps your file name with -redacted added.

• Open the file and tick what to look for
• Press ‘Find and mark’ and read every page
• Draw boxes for misses, press × on wrong ones
• Press ‘Redact and save’, then ‘Download’

Cases the automatic search tends to miss

A scanned page is a photograph, so its characters cannot be read automatically and nothing gets marked there. Draw the boxes yourself on those pages. A name or address broken across two lines may not be joined up either, and phone numbers come in so many styles that they are the least reliable category.

Numbers split into separate table cells can also slip through because they no longer look like a single number. For that reason it is wise to go through the whole file once more after the automatic marking, zoomed in, from the first page to the last.

What changes in the file afterwards

Every marked page becomes a picture, so its text can no longer be selected or searched, and links and form fields on it disappear. Because pictures take more room, the file can grow. Pages you did not mark keep their text and layout as they were.

The original is untouched and a new copy is made, so keep the original somewhere safe. Share only the redacted copy, and use clearly different file names so the unredacted one never goes out by mistake.

The check to run on the saved file

Open the redacted copy as a new file and drag across the boxed areas. Nothing should be selectable. Then search for the first digits of a number you covered; a clean file returns no result.

Make this a habit before attaching anything to an email. A number you thought you had hidden is often repeated in a header, a footer or another table cell on a different page.

If the office you submit to sets a format for copies, check against that too. If the redacted copy has grown large because of the picture pages, you can run it through Compress PDF before sending.

Where the work happens

Finding and redacting both happen in your browser and the file is not uploaded, which matters when the document holds personal data. To let other tools open the same file, the page keeps a copy of the PDF you chose in your browser’s own storage for up to two hours, and that copy stays on your device.

The tool cannot decide what a given office legally requires you to hide. If the form lists specific items, read that list first and compare each item against the finished file.

When handling several people’s paperwork, redact one person’s file at a time and save each under a distinct name. That prevents information meant to be hidden in one file from slipping into another’s.

A standard to settle before you submit

What has to be hidden differs by recipient. Some ask only for the last digits of an ID number to be covered, while others also want account numbers or addresses. Read the instructions first, make a list and mark the file against that list.

A file that has been sent cannot be taken back. Opening the redacted copy once more before you press send, and checking the file name to be sure you attached it and not the original, is the most effective safeguard.