Redact PDF

Draw boxes over what must go and the marked pages are rebuilt as flattened images, so the text underneath is gone rather than hidden. No upload.

🌐 Español

Redact PDF works entirely in your browser — upload a PDF above, drag black boxes over the content you need to remove, and download a version where each boxed page has been rebuilt as a flat image so the hidden text is genuinely gone. Your document never leaves your device.

🔒 Private by design: everything runs locally in your browser and never uploaded to any server.

Written and maintained by Last updated

The black rectangle problem

There is a specific, recurring failure that has embarrassed newspapers, law firms and government departments, and it comes from one misunderstanding about how PDFs work.

A PDF is not a picture of a page. It is a list of objects: text runs with positions and fonts, vector paths, embedded images. Drawing a filled black rectangle adds one more object to the top of that list. It hides the text visually and changes nothing about the text itself.

So the words are still there. Select across the box and you get them. Run a text extraction tool and you get them. Open the file in an editor and delete the rectangle and there they are. Every published incident of this kind has followed the same pattern, and the pattern is not going to stop while editors offer a rectangle tool next to a redaction label.

This tool takes the only approach that actually works in a browser: it throws the page away and rebuilds it from pixels.

Redacting a document

  1. Click Upload a PDF to redact and choose your file. The document is rendered page by page in the browser.
  2. Use Previous page and Next page to move through it. The indicator between them shows where you are.
  3. Click and drag on the page to draw a black box over anything that must go. Draw as many as you need.
  4. Undo on this page removes the last box you drew here. Clear this page removes them all.
  5. Watch the hint under the controls. It tells you how many pages will be flattened, and the download stays disabled until you have drawn at least one box.
  6. Click Download redacted PDF. The button changes while the export runs.
  7. Open the result and try to select the text on a redacted page. There should be none.

That last step is worth doing every single time. It takes three seconds and it is the only direct confirmation that the redaction worked.

What flattening actually costs

Being honest about the trade is more useful than pretending there is not one.

When a page is marked, the whole page is rendered to a bitmap, the marked regions are painted over so those pixels are destroyed in the bitmap, and the output page contains that image and nothing else. Every piece of text on that page, including everything you did not box, is now part of an image.

The consequences are real. That page is no longer searchable. Its text cannot be selected or copied. A screen reader cannot read it. It will not reflow on a small screen. And it is likely to be larger in bytes than the vector page it replaced.

Those costs are confined to the pages you marked. Every other page is copied through as the original vector PDF, keeping its text layer, its quality and its size. That is why marking three pages of a long report costs you almost nothing, and marking every page costs you a great deal.

Rendering happens at 150 dots per inch, which keeps text readable on screen and in normal printing without making the file enormous. It is lower than the vector original, so very small print will look a little softer.

What redaction does not cover

Removing content from the page is not the same as removing information from the file.

Document metadata is handled here: the output is assembled as a fresh document, so the original title and author do not carry across. But other things can survive a page level redaction, including attachments, form field values and annotation authors, none of which live in the page content stream.

If your concern is authorship and provenance rather than content, run Remove PDF Metadata as well, which strips both metadata stores including the XMP block that most cleaners miss.

And remember the obvious one: a redacted document that is a scan of a printed page may still show an impression through the paper, or a name in a header you did not box. Read the output, do not just check that the boxes are there.

Flatten PDF solves a neighbouring problem, turning a filled form into static page content so the fields can no longer be edited. That locks down input rather than removing content, so it is not a substitute for redaction.

Protect PDF adds a password, which controls who can open a file rather than what is in it, and the two solve different problems. Sign PDF adds a signature, and Annotate PDF draws boxes and notes on a page while deliberately leaving the text underneath intact, which is exactly the behaviour this tool exists to avoid. The rest is on the PDF tools hub.

See it in action

Screenshot of the Redact PDF tool with an empty dropzone waiting for a PDF to redact
Redact PDF mid-process: an empty dropzone waiting for a PDF to redact.
Screenshot of the Redact PDF result screen showing the PDF rendered page by page, ready for redaction boxes to be drawn over the parts to remove
The finished result: the PDF rendered page by page, ready for redaction boxes to be drawn over the parts to remove. The download link is a local blob URL — the file never leaves your device.

Frequently asked questions

Why is a black rectangle in a normal PDF editor not redaction?

Because a PDF keeps its text as text and its drawings as separate objects layered on top. Adding a black rectangle adds one more object above the words; it does not remove them. Anyone can select through it, copy the text, extract it with a command line tool, or open the file in an editor and delete the box. Newspapers and government departments have published documents this way and had them read within hours.

What does this do instead?

Destroys the pixels. Each page you mark is rendered to a canvas, the marked regions are painted opaque black on that canvas so those pixels no longer exist, and the output page is rebuilt from that flattened image alone. The page's original text and vector objects are never copied into the result, so there is nothing left to extract.

What is the cost of that approach?

A marked page loses its entire text layer, not just the boxed part. Everything else on that page becomes part of the image too, so it is no longer selectable, searchable or reflowable. That is the unavoidable price of the guarantee, and it is why only the pages you actually marked are flattened while every other page is copied through as the original vector PDF.

Does the file get bigger?

On the redacted pages, usually yes, because an image of a page of text is larger than the text was. Pages you did not mark are untouched, so a two hundred page report with three redacted pages grows only slightly. A document where every page is marked will grow substantially.

What resolution are the flattened pages?

They are rendered at 150 dots per inch, which is a deliberate middle setting. It is high enough that text stays comfortably readable on screen and in ordinary printing, and low enough that a long document does not become enormous. It is noticeably lower than the underlying vector page, so very fine print will look softer than it did.

Does the document metadata come across?

No. The output is built as a fresh document with pages copied or embedded into it, so the original file's information dictionary, including its title and author, does not carry over. That is verified by a test rather than assumed, and it removes one of the ways a redacted document can still identify its source.

Can I redact just some pages?

That is the default behaviour. Only pages carrying at least one box are flattened; the rest are copied through unchanged as the original vector PDF, keeping their quality and their searchable text. The hint under the controls tells you how many pages will be flattened before you export.

Is my document uploaded?

No, and for this tool that is the whole point. Rendering, painting and rebuilding all happen inside your browser tab. A document that needs redaction is by definition a document containing something you do not want other people to see, so handing it to a server first would be a strange way to begin.

Related tools