📄 PDF · Updated October 16, 2026 · 8 min read

Redact a PDF Permanently: Find, Mark and Remove Text

To redact a PDF permanently, mark every sensitive string, burn solid boxes into the page, and save a new file in which the words under those boxes no longer exist. GrabCast's free Redact PDF does that in your browser: it finds Social Security numbers, emails, US phone numbers, card numbers and dates in one click, lets you draw boxes over anything else, and checks the saved copy to prove the marked text is gone. This guide walks through a real example, a fictional two-page new hire record, from opening the file to verifying the output.

⬛ Try Redact PDF now — freeOpen →

Redact PDF toolbar reading 8/8 matches on, 1 page to redact, above page 1 of the new hire record with the SSN, email and phone fields covered — Personal identifiers found and removed from a fictional HR record in under a minute.

💡 Why a black rectangle is not enough

Most PDF viewers let you draw a filled shape or highlight in black. That shape is an annotation sitting on top of the page, and the original characters stay in the content stream underneath. Anyone can select the area, copy it into a text editor, or delete the annotation and read everything. Court filings, HR files and insurance claims have leaked exactly this way. Real redaction removes the underlying text, images and metadata, so there is nothing left to uncover.

What happens when you redact a PDF with GrabCast

The tool opens your file with PDF.js, the same engine Firefox uses to display PDFs, and reads the text layer of every page. When you export, each page that carries at least one mark is rendered to an image at the quality you choose, the boxes are painted into those pixels, and that image replaces the original page in a brand-new PDF built with pdf-lib.

Because the new page is a picture, there is no text object left for the redacted words to hide in. Pages without any mark are copied across unchanged, so in a 30-page contract where only page 4 mentions a salary, the other 29 pages keep selectable, searchable text.

Metadata is dropped: title, author, subject, keywords, creator, producer and XMP data do not carry over to the new file.
Bookmarks and file attachments are removed, since either can repeat names or embed an unredacted copy.
Form fields are flattened into the page, or removed entirely if you tick the option to strip annotations, comments, links and fields.
Links that point to other pages are dropped so no stray reference pulls hidden content along.

After saving, the tool reopens the result and searches it for every string you marked. In our test it reported that none of the 8 marked strings could be found and that the redacted page contained no text at all.

Finding sensitive data before you redact the PDF

Manual scanning misses things, especially in dense forms. The one-click detectors work on the text layer and list every hit with its page number so you can review them in one place.

Emails: any address pattern, such as jordan.lee@example.com in our sample record.
US phone numbers: formats like (555) 014-2290, 555-014-2290 or +1 555 014 2290.
SSNs: nine digits written as 123-45-6789 or with spaces between the groups.
Card numbers: 13 to 19 digits that pass the Luhn checksum, which filters out most invoice and order numbers.
Dates: 01/15/2024, 2024-01-15 and January 15, 2024 styles, useful for birth dates.
Your own terms: type a name, account label or project code, and choose Whole word or Ignore case.

On the sample record the three buttons found two SSNs, three email addresses and three phone numbers, eight items in total, including a spouse SSN buried in a paragraph that is easy to skip when you skim. Untick anything that should stay visible, such as a company switchboard number you are happy to publish.

Choosing quality, format and box style

Since redacted pages become images, the export settings decide how readable they stay and how big the file gets.

200 DPI JPEG is the default and suits most office documents; our two-page sample came out at 0.31 MB.
300 DPI with PNG keeps small print and thin table lines crisp, at the cost of a larger file.
150 DPI is fine for plain letters that will only be read on screen.
Box color defaults to black; a label of up to 40 characters, such as REDACTED or the legal reason code your office uses, is printed on every box.

The on-screen marks are deliberately translucent so you can see what sits underneath and adjust them. The exported boxes are solid, and the text below them is not in the file at all.

Limits worth knowing before you share a redacted PDF

Scanned pages have no text layer, so search and the detectors cannot see them. The tool warns you which pages have no selectable text; draw boxes over those by hand, or use Redact whole page.
Redacted pages lose selectable text. If a recipient needs to search them, run the output through OCR PDF afterwards; the covered words will not come back, because they are no longer in the image.
Files over 200 pages need to be split first, and password-protected files must be unlocked with the password you already know.
Detectors follow patterns, not meaning. A name, a street address or a medical condition only gets marked if you search for it or draw a box.

Keep the original somewhere safe and send only the copy ending in -redacted.pdf. Open that copy, press Ctrl+F or Cmd+F, and search for one of the removed values yourself before it leaves your hands.

After redacting, lock the file with the Protect PDF tool before you share it.

Step-by-step

1Open the PDF — Drop the file into Redact PDF; every page renders on your device and nothing is uploaded.

Redact PDF with new-hire-record.pdf open: search box, Find and mark button and one-click Emails, US phone numbers, SSNs, Card numbers and Dates buttons — Open the PDF; the pages render on your device and the one-click detectors sit above them.

2Find the data — Click SSNs, Emails, US phone numbers or the other detectors, type any names, then review the checklist and untick items to keep.

Found matches list: SSNs 2/2 (123-45-6789, 987-65-4321), three example.com email addresses and three 555 phone numbers, all checked — Click SSNs, Emails and US phone numbers; every hit lands in a checklist you can review and untick.

3Check the marks — Scroll through the page previews, drag extra boxes over signatures or photos, and confirm the count reads the pages you expect.

Page 1 of the new hire record with dark boxes over the SSN, email and phone fields while name, position and start date stay visible — The marks appear on the page preview, so you can see exactly what will be blacked out.

4Export and verify — Pick quality and format, click Redact & download PDF, and read the Verified line before you send the -redacted copy.

Export options set to 200 DPI JPEG, then the status line: 1 page redacted, 1 page unchanged, metadata removed, and a green Verified check — Click Redact & download PDF; the tool reopens the new file and confirms the marked text is gone.

Common mistakes to avoid

⚠️Drawing a black rectangle or highlight in a regular PDF editor and assuming the words underneath are gone; they can still be copied.

⚠️Relying on the detectors alone and forgetting names, addresses or account labels that only a text search or a hand-drawn box will catch.

⚠️Skipping scanned pages, where search finds nothing because there is no text layer, even though the page clearly shows a number.

⚠️Sending the original by accident; always attach the file whose name ends in -redacted.pdf and search it once yourself.

Pro tips

✓Search for the person's name in both orders, such as Jordan Lee and Lee, Jordan, because forms often print it both ways.

✓Use the Show link next to each match to jump to it on the page and confirm the box covers the whole value.

✓Add a short label like REDACTED so readers know text was removed on purpose rather than lost in printing.

✓Choose PNG at 300 DPI for tables with small numbers, and JPEG at 200 DPI for ordinary letters to keep the file small.

✓Tick the option to remove annotations and form fields when a file has comments or filled fields you have not reviewed line by line.

Frequently asked questions

Can someone recover text I redacted with this tool?

Not from the redacted pages. Each one is rebuilt as an image with the boxes painted into the pixels, and the original text, fonts and objects of that page are not copied into the new file. The tool then searches the result to confirm the marked strings are absent.

Is my PDF uploaded anywhere?

No. PDF.js displays and searches the pages and pdf-lib writes the new file, both inside your browser tab. The document stays on your computer or phone.

Why is the redacted file bigger than the original?

Text is compact, while a page image at 200 or 300 DPI takes more bytes. Choose JPEG and 150 or 200 DPI for a smaller file, or redact only the pages that need it so the rest stay as compact text.

Does it find names automatically?

No. Names have no fixed pattern, so type them into the search box. The one-click detectors cover emails, US phone numbers, SSNs, card numbers and dates.

Can I redact a scanned PDF?

Yes, by drawing boxes or using Redact whole page. Search cannot see scanned pages because they carry no text layer, and the tool tells you which pages those are.

📌 Bottom line

Real redaction means the covered words are gone from the file, not hidden behind a shape. Use the detectors and a name search to find everything, draw boxes for the rest, export, and trust the Verified line only after a quick search of your own in the -redacted copy.

Open Redact PDF →