How to Safely Redact Legal Documents in Microsoft Word

Word LX guide to redacting legal documents in Microsoft Word

A paralegal blacks out a client’s account number in a settlement agreement. The bar looks solid on screen. The PDF goes to opposing counsel, who selects the page, presses Ctrl+C, pastes into a text editor, and reads the number. That takes about 4 seconds and needs no special tools.

If you want to redact legal documents in Microsoft Word without that happening, start with the fact almost nobody is told: Word has no redaction feature. Black highlighting, white font and a drawn black rectangle are formatting effects layered over text that stays in the file.

The short answer

To redact legal documents in Microsoft Word safely, delete the sensitive text and replace it with a marker such as [REDACTED]. Run Document Inspector to strip comments, tracked changes and metadata. Convert to PDF, then flatten the file. Word has no redaction tool, so black highlighting and black shapes leave the original text recoverable.


Table of contents


Does Microsoft Word have a redaction tool

No. There is no redact command in Word for Microsoft 365, Word for the web, or any perpetual version. Microsoft did publish a Word 2007 Redaction Tool add-in that genuinely deleted the marked text and dropped a black bar in its place. It lived on CodePlex, and Microsoft shut CodePlex down on December 15, 2017. The add-in went with it. Unsupported copies still circulate on third party download sites.

2 Microsoft products do redact, and neither one is Word. Microsoft Purview applies sensitivity labels that encrypt a file and control who can open it. Azure AI Language offers document PII redaction as a cloud service you call against an uploaded file. Inside Word itself, every option on the ribbon is a formatting change.


Why covering text in Word does not remove it

A Word file keeps text and formatting in separate places. Highlighting, font colour and drawn shapes are instructions about how a character should look. The character sits in the document body regardless of what is layered on top of it.

The NSA put this in writing 20 years ago. Its guidance for federal agencies, Redacting with Confidence (report I333-015R-2005, December 2005), says covering text with black rectangles or highlighting text in black is “a common and effective means of redaction for hardcopy printed materials” and that “it is not effective, in general, for computer documents distributed across computer networks.”

Printing to PDF leaves the problem in place. The same guidance warns that “merely converting an MS Word document to PDF does not remove all metadata automatically.”


What each fake redaction method leaves behind

The 5 ways people redact legal documents in Microsoft Word that leave the original text in the file
All 5 of these leave every original character sitting in the document.

5 methods come up over and over in legal documents. All 5 leave every original character in the file.

What you do in Word What it looks like What is still in the file How it comes back out
Black highlight over the text A solid black bar Every original character Select all, paste into a text editor
Font colour changed to white Blank white space Every original character Select all and paste, or set the colour back
Black rectangle drawn over the text A solid black box Every original character Drag the shape aside, or copy the text beneath it
Text box with a black fill A solid black block Every original character Same as the rectangle
Hidden font effect applied Nothing visible Every original character Toggle formatting marks, or enable print hidden text
Text deleted and replaced with a marker [REDACTED] Nothing Nothing to recover

Pixelating or blurring belongs in the first group. Dan Petro, lead researcher at Bishop Fox, built a tool that reverses pixelated redactions and warned in 2022 that “the last thing you need after making a great technical document is to accidentally leak sensitive information because of an insecure redaction technique.”


How a failed redaction gets found

Recovery takes no forensic tooling. Any of these works on an ordinary laptop.

  1. Select the whole page, copy, and paste into a plain text editor. The hidden text arrives with everything else.
  2. Drag the black shape to one side. The text is sitting underneath.
  3. Search the document for a name or a number you expect to be gone. Find reads the text layer, not the picture on screen.
  4. Run text extraction on the PDF with any free utility.

The Manafort filing in January 2019 went through option 1. Reporters selected the blacked out passages in a court PDF, pasted them elsewhere, and read the material his lawyers had tried to withhold.


What the court record says about failed redactions

2 studies measured this at scale, and both looked at federal filings.

The Federal Judicial Center searched a single month of filings, November 2013, and found 16,811 instances of unredacted full Social Security numbers across 5,437 documents, affecting 5,031 people. At least 314 of those were failed attempts at redaction. The report names the cause: “the apparent use of word processing redaction techniques that retain the Social Security number in the metadata when the document is converted to PDF for filing in court.”

Princeton’s Center for Information Technology Policy scanned 1.8 million PACER documents in 2011. Roughly 2,000 carried black redaction rectangles, and 194 of those had recoverable text underneath. Close to 1/10 attempted redactions failed. If you want more context on how Word behaves in legal work, we pulled together 3 statistics every legal professional should know.


3 redactions that failed in real filings

United States v. Manafort, January 2019. Defence counsel filed a response in the District of Columbia with black boxes over key passages. The boxes were unflattened overlays. Copying and pasting revealed that Manafort had shared 2016 campaign polling data with Konstantin Kilimnik.

Apple v. Samsung, December 2011. A sealed opinion from Judge Lucy Koh in the Northern District of California was posted with redactions that copy and paste defeated, exposing Apple’s licensing terms with Nokia and IBM. The court pulled the document and reposted a corrected version within hours.

The TSA screening manual, 2009. The Transportation Security Administration posted its standard operating procedures with sections blacked out. Anyone who pasted the document into Word or Notepad could read the hidden sections, including screening exceptions and sample credentials.


How to redact legal documents in Microsoft Word safely

The 5 step method that removes text instead of covering it
Delete, inspect, convert, flatten, verify. The order matters.

The method the authorities agree on has 5 steps, and the order matters.

  1. Work on a copy. Keep the unredacted original in your document management system.
  2. Delete the text. Select it and remove it. Type [REDACTED] in its place so the reader can see something was taken out.
  3. Run Document Inspector. File, Info, Check for Issues, Inspect Document. Clear comments, revisions, document properties, personal information, hidden text and custom XML.
  4. Convert to PDF.
  5. Flatten the PDF. The UK National Archives Redaction Toolkit tells reviewers to round trip the PDF through an image format so no text layer survives. Acrobat’s Sanitize Document does the same job.

Drawing a shape and picking a colour appear nowhere in that list. If your team is building broader Word discipline, our 9 best practices for lawyers using Microsoft Word covers the habits that sit around this one.


How to verify the redaction actually worked

3 checks, about 60 seconds in total, run on the final PDF rather than the Word file.

Open the PDF, press Ctrl+A, copy, and paste into a plain text editor. Read what comes out. Everything you redacted should be absent.

Search the PDF for a term you removed. A name, an account number, a dollar figure. Search reads the text layer, so a hit means the text survived the process.

Check the file properties for author, company and title fields.

The Department of Homeland Security wrote this into policy in 2010, requiring “an independent quality control check on the redacted material to verify and assure that permanent redaction has been successful.” A second person catches what the first one stopped seeing.

Free download

The Word Redaction Pre Send Checklist

One page. The 5 step method, the 3 verification tests, and the 10 point check to run before any redacted document leaves the firm. Print it and tape it to the wall.

Download the checklist (PDF)

What Document Inspector removes and what it misses

Document Inspector handles comments, revisions, versions and annotations, document properties and personal information, custom XML data, headers, footers and watermarks, invisible content, and hidden text. On metadata it does a real job.

It misses content that is only visually hidden. Microsoft says so in its own documentation on what Document Inspector doesn’t catch: white text on a white background, a shape coloured to match its background, or a shape stacked over other content. The tool “sees it and thinks it’s just normal data.”

That is the whole problem in one line. The inspector has no way to flag your black box, because as far as Word is concerned you drew a rectangle, and drawing a rectangle is a legitimate thing to do.


The metadata half of the problem

Redaction deals with what a reader can see. Metadata deals with everything else the file carries: who wrote it, who edited it, when, what the earlier wording was, and which comments the team left for each other.

Tracked changes and comments are the sharpest risk in a legal document. An accepted change is gone. A change still pending sits in the file with both versions of the language, which shows opposing counsel exactly what you were prepared to concede.

ABA Formal Opinion 06-442, issued August 2006, notes that lawyers sending documents can limit the risk by scrubbing metadata or by sending a version without the embedded information. We covered the other half of this problem in why removing metadata is good for brand consistency, which is worth reading alongside this one.


Where else content hides in a Word file

Cropped images catch people out most often. Cropping in Word hides part of a picture and keeps the rest. Microsoft’s own documentation states that cropped parts “are not removed from the file, and can potentially be seen by others”. Compress Pictures with “Delete cropped areas of pictures” ticked is what removes them.

Then there are the quieter places. Headers and footers carry matter numbers and internal file paths. Embedded spreadsheets carry every row, including the ones outside the visible range. Text boxes dragged off the edge of the page stay in the file. SmartArt and charts keep their source data. Recycling an old document brings all of it along, which is why reusing old legal documents is dangerous.


What the rules require you to redact

Federal Rule of Civil Procedure 5.2 sets the floor for civil filings. 4 categories: Social Security and taxpayer identification numbers cut to the last 4 digits, a birth date cut to the year, a minor’s name cut to initials, and a financial account number cut to the last 4 digits.

Federal Rule of Criminal Procedure 49.1 adds a fifth category for criminal filings. A home address is cut to city and state.

Both rules put the work on the filer, and the Administrative Office of the US Courts says the same in its privacy policy for electronic case files. The clerk does not check your redactions, and a filing that leaks is your filing.


Why redaction keeps failing at firm level

Look at when redaction happens. It is the last thing before a document goes out, done against a filing deadline, usually by whoever has the file open. There is no standard step, no checklist, and no second pair of eyes.

The tooling gap is real too. Craig Ball, a board certified computer forensics examiner and e-discovery special master, put it plainly in 2019: “Isn’t it time we started demanding tools to redact natively instead of clinging to the idiotic, error-prone and labor-intensive ways we do it?”

Firm size shapes the exposure. ILTA’s 2024 technology survey data, reported by Litera, puts metadata cleaning tool adoption at 26% in firms under 50 lawyers against 73% in firms over 700.


Making redaction part of the document process

A firm gets consistent redaction the same way it gets consistent numbering and consistent styles. The rule lives in the document rather than in one person’s memory.

In practice that means 3 things. A standard place where confidential fields sit, so the person redacting knows what to look for. Templates that keep matter numbers and internal file paths out of headers and footers where they get forgotten. And a pre send check that someone has to complete rather than a habit some people have.

Word LX governs the document layer inside Microsoft Word, which is where this risk lives. Firms that set law firm document standards and hold them are the ones where a last minute redaction is a controlled step instead of an improvisation. The same logic applies to Microsoft Word styles for lawyers.


A 60 second pre send check

Before any redacted document leaves the firm:

  • The sensitive text was deleted, not covered
  • Document Inspector was run and came back clean
  • Tracked changes were accepted or rejected, with none left pending
  • Comments were deleted
  • Cropped images had their cropped areas deleted
  • Headers and footers were checked for matter numbers and file paths
  • The file was converted to PDF and flattened
  • Select all, copy, paste into a text editor returned nothing sensitive
  • A search for 2 removed terms returned 0 hits
  • A second person confirmed all of the above

Redacting legal documents in Microsoft Word FAQ

Does Microsoft Word have a redaction tool?

Microsoft Word has no redaction tool in any current version, including Word for Microsoft 365 and Word for the web. Microsoft’s Word 2007 Redaction Tool add-in was retired when CodePlex closed in December 2017. To redact legal documents in Microsoft Word you delete the text yourself, then convert to PDF and flatten the result.

Is black highlighting in Word a real redaction?

Black highlighting in Word is not a real redaction. Highlighting changes how a character is displayed and leaves the character itself sitting in the document. Anyone can select the page, copy it, and paste it into a plain text editor to read everything under the black bar. White font and drawn shapes fail the same way.

Does saving a Word document as a PDF remove redacted text?

Saving a Word document as a PDF does not remove text you covered rather than deleted. NSA guidance warns that converting Word to PDF does not remove all metadata automatically, and a black shape carried into the PDF still has selectable text underneath it. Deletion has to happen in Word first.

Can redacted text be recovered by copy and paste?

Redacted text can be recovered by copy and paste whenever the original characters remain in the file. Select the page, copy, paste into a plain text editor, and the hidden text appears. This is what happened in the January 2019 Manafort filing and in a sealed 2011 Apple v. Samsung opinion.

What does Word’s Document Inspector remove?

Word’s Document Inspector removes comments, revisions and annotations, document properties and personal information, custom XML data, headers, footers and watermarks, invisible content, and hidden text. It leaves text you covered with a shape or coloured white, because Microsoft says the tool treats visually hidden content as ordinary data.

How do you check a redaction worked before you file?

Check a redaction before filing by opening the final PDF, selecting all, copying, and pasting into a plain text editor to see what comes out. Then search the PDF for 2 terms you removed and confirm 0 hits. Have a second person repeat both checks independently.

Control what leaves the firm

Make the safe version the easy version

Book a walkthrough and we will look at where confidential content sits in your templates, your headers and footers, and your document workflow, then show you what Word LX governs inside Microsoft Word.

Book a Word LX demo