Proper redaction removes the text and image data under the area, then usually rasterizes that area so nothing can be recovered. Drawing a box, changing text to white or deleting with a comment tool only covers it. After redacting, search the file for the words you removed and try selecting text under each box.
Redaction failures are rarely clever attacks. They are usually a rectangle drawn on top of text that is still in the file, where anyone can copy it out. The Northern District of California’s e-filing page is one of the clearest public explanations, because courts see the consequences.[1]
01The wrong ways, per the court
Under “Wrong Ways To Redact”, the court lists methods not to use:[1]
- Changing the font to white. “Turning the text white makes the words seem to disappear, but they don't.”
- Blacking out with comment tools. “Edits made by graphic and "commenting" tools which can black-out, cover over or remove sections of text can still be removed by anyone to reveal the text underneath.”
- Deleting in a word processor. “Metadata can reveal anything that was contained in the file at any time, even text that was previously deleted or changed, and even if the file was re-saved.”
- Covering paper with marker or tape, then scanning. “A scanned document redacted this way may still provide enough image detail to enable someone to see what was assumed hidden”
If you can select it, search for it or copy it, it was never redacted. The test is what is in the file, not what you see.
Our reading of [1]
02What real redaction does
A PDF page is drawn from instructions: “put these characters here in this font”, “draw this image here”. A box added on top is just one more instruction, drawn last. True redaction edits the instructions underneath: it deletes the text objects inside the area, removes or blanks the image pixels inside it, and then often turns the area into a flat image so no outline of the original survives. It should also clear hidden copies: metadata, bookmarks, form field values and attached files can repeat the same words.
The court’s own preferred method avoids the problem at the source: replace the sensitive words in the original document before it ever becomes a PDF, so “the source contains no unwanted text or data to begin with”.[1]
03What courts require you to redact
In federal civil cases, Rule 5.2 says that unless the court orders otherwise, a filing that contains these items “may include only”:[2]
- “the last four digits of the social-security number and taxpayer-identification number;”
- “the year of the individual's birth;”
- “the minor's initials; and”
- “the last four digits of the financial-account number.”
Criminal filings follow Rule 49.1, which covers the same items and adds one: a filing may include only “the city and state of the home address”.[3]
The court page adds other information that is often appropriate to redact, including “Medical information”, “Trade secrets” and “Informant names”.[1] Outside court, the same discipline applies to anything you share: a tax return sent to a landlord or a bank statement uploaded to an AI assistant.
04Check your redaction in two minutes
- Open the redacted PDF and use Find (Ctrl+F or Cmd+F) for each word or number you removed. Nothing should match.
- Drag to select across each black box, copy, and paste into a plain text editor. Nothing from under the box should appear.
- Check document properties for the title, author and subject fields, and remove metadata if they carry anything sensitive.
- Or drop the file on Check redaction, which looks for text that is still present under dark rectangles, in your browser.
Redact PDF on pdfnex removes the text under each area and rasterizes it, can find SSN, EIN, email and phone patterns for you, and runs the check on the output. The file never leaves your browser, which matters most for exactly the documents that need redacting.
This guide explains redaction methods and quotes a federal rule and a court’s guidance. It is not legal advice; follow your court’s rules and any order in your case.