Redact a document or photo

Draw over what should not be seen. The marks are burned into the pixels, so what was underneath is gone rather than hidden. Everything happens in this browser tab.

How it works

A black rectangle drawn on top of a PDF is not a redaction. It is a shape sitting above text that is still there, still selectable, still returned by any text extractor and still in the file when someone copies the page into an email. This has embarrassed newspapers, law firms and governments, and it keeps happening because the document looks right on screen. The failure is invisible until it is catastrophic.

This tool cannot fail that way, because it never draws on the document. Each page is rasterised first, which turns it into pixels and discards every text object, font reference and vector path it ever contained. The boxes are then painted onto those pixels, and a new document is built from the result. There is no layer underneath, because there is no longer any layer at all.

The price is the thing you should decide about consciously: the output has no selectable text. It cannot be searched, and a screen reader cannot read it. For a document you are sending because someone asked for specific information to be removed, that is usually the right trade and often the point. For a contract that people still need to search, it is the wrong tool, and you would want the original with the sensitive parts removed at the source instead.

Photos work the same way. The pixels under each box are overwritten rather than hidden, and re-encoding drops the metadata as a side effect, so the copy no longer carries the GPS coordinates of where it was taken. Nothing is uploaded at any point, which matters more here than almost anywhere else on this site: the files people redact are exactly the ones they cannot afford to hand to a stranger.

How to use it

  1. 1 Choose a PDF or a photo, or paste one straight in.
  2. 2 Drag a box over anything that should not be seen. Add as many as you need, on as many pages as you need.
  3. 3 Press Redact and download.
  4. 4 Open the copy and try to select the text under a box. There will be nothing to select.

Good to know

  • Boxes are burned in, not drawn on top. What was underneath is destroyed.
  • A redacted PDF becomes a document of pictures, so it is no longer searchable. That is the cost of doing this properly.
  • To delete pages entirely rather than parts of them, use the PDF splitter and keep the pages you want.
  • The redacted copy carries no metadata from the original, including the author of a PDF and the location of a photo.
  • Check the result before you send it. It takes five seconds and it is the whole point.

Browser support

  • Chrome 94 and later
  • Firefox 93 and later · full speed from 105
  • Safari 16.4 and later
  • Edge 94 and later

These are the oldest versions with everything the tool uses. They are worked out from the tool itself, so they change when it does.

Version numbers checked on 2026-09-04.

The black rectangle that is only a drawing

In 2011 a court filing about a well known technology dispute was published with sections blacked out, and within hours people had the text: the blocks were rectangles drawn over a page whose words were still underneath, and selecting the region and pasting it revealed everything. The same mistake has been made since by government departments, law firms and newspapers, and it will be made again next month. It is not a rare accident. It is the default outcome of the most obvious way to do the job.

The reason is structural. A PDF is a collection of objects: text-showing operators with their fonts and positions, images, vector paths. Drawing a filled rectangle adds one more object to that collection, on top in painting order. It changes what a viewer displays and nothing about what the file contains. The text operators are untouched, so text extraction returns them, copy and paste returns them, a screen reader reads them aloud, and any of a hundred libraries can list them in a line of code. The document looks redacted to a human and is not redacted at all to a program.

Genuine redaction has to remove the objects, and there are two ways. The precise one is surgical: parse the content stream, find the text-showing operators whose glyph positions fall inside the region, and rewrite the stream without them. It preserves the rest of the page as real text and it is genuinely difficult, because glyphs are positioned by a running matrix that later operators depend on, fonts may be subset so removing the last use of a glyph can invalidate the font, and a partially covered word has to be split rather than dropped.

The blunt one is to rasterise. Render the page to pixels at a sensible resolution, paint the boxes onto those pixels, and build a new document with the result as an image. Every text object in the original is gone, not because it was carefully removed but because nothing from the original structure survived the trip through a bitmap. There is no content stream left to mine. This is what the tool does, and the honesty of it is the point: it is impossible to get subtly wrong, which is exactly the failure mode the surgical approach has.

The trade is real and worth stating plainly. The output cannot be searched, selected or read aloud, and it is larger. A redaction you can still search is a redaction that still contains the text, so those two properties cannot both be had for the same words. What you can have is a document where the words that mattered are unrecoverable, which is the requirement people actually have when they reach for this. Rounding also matters more than it sounds: the boxes are rounded outward to whole pixels, because a box that rounds inward can leave a one pixel line of the original along its edge, and a redaction that leaks a sliver of what it hid is worse than none, since it looks finished.

Frequently asked questions

How do I redact a PDF so the text is really gone?

Choose the file here and draw over what should be removed. The page is rasterised before the boxes are applied, so the underlying text objects no longer exist in the output. Try selecting the text in the downloaded copy: there is nothing there to select.

Why can I still select text in redactions from other tools?

Because most of them draw a filled rectangle over the page and save it. The rectangle is a drawing; the text is a separate object underneath, unchanged. Anything that reads the file rather than looking at it will find the words, which is how redaction failures reach the news.

Will my redacted PDF still be searchable?

No, and it cannot be. Making the text unrecoverable and keeping the text searchable are the same question with opposite answers. If you need a searchable document, remove the sensitive material in whatever produced the file, then export it again.

Can I redact a photo?

Yes. Choose an image instead of a PDF and draw over faces, plates, addresses or anything else. The pixels under each box are overwritten and the file is re-encoded, so the original values are not recoverable from the copy.

Is the file uploaded?

No. Rendering, painting and saving all happen inside this browser tab on your own device. You can confirm it in your browser’s developer tools on the Network tab: nothing carrying the document goes out. That matters here more than anywhere, because the documents people redact are the ones they cannot share.

Does the redacted copy keep the original metadata?

No. The output is a new document built from images, so it carries no author, title, producer or creation date from the original, and a redacted photo carries no EXIF. Metadata is not covered by any box, so removing it is part of the job.

Related tools

How this tool was built — a step-by-step account of the engine, the code and what went wrong.

Your files are processed on your device and are not uploaded. You can verify this in your browser’s developer tools under the Network tab.