Skip to content

Redact PDF

Black out sensitive text and remove it from the file.

Drop a PDF to redact

Only your browser reads them.

How to use Redact PDF

  1. Open your PDF. Drop a file onto the page or choose one from your device to view page 1.
  2. Mark sensitive areas. Drag across text, signatures, or photos, or use Find and mark text with Email addresses, Phone numbers, or Long numbers.
  3. Choose cleanup options. Keep Also remove document properties and Also remove bookmarks and page labels checked to strip hidden metadata.
  4. Apply and download. Select Redact to burn black boxes into the affected pages and save the sanitized PDF.

Your files are processed on this device and never uploaded. They stay private, and there's nothing to delete afterwards.

Why drawing a black box over PDF text fails as redaction

Covering words with a dark rectangle in a standard editor only paints a shape on top of existing glyph operators. Anyone opening the file can still copy, search, or extract the original characters directly from the underlying PDF content stream.

How page-level rasterization eliminates underlying text streams

When you apply redactions in this tool, your browser renders each marked page onto an HTML5 canvas, paints solid black pixels with a 1 px padding margin over every box, and replaces the original page with that flat image. Because the old page object and its text operators are removed, no software can select or copy hidden characters underneath.

Under the PDF Association ISO 32000-2 PDF 2.0 specification, visual annotations and page content streams are stored as separate dictionary objects. After swapping each redacted page, the worker runs an unreachable-object prune so orphaned text streams do not linger inside the saved file.

Preserving vector text on untouched pages

Flattening an entire 50-page contract into images just to hide one social security number on page 2 would inflate file size and ruin text searchability elsewhere. Instead, pages without redaction marks are copied forward intact.

After saving your file, you can verify that no hidden text remains under black boxes across every page. Running an automated check confirms that no covered glyphs or unapplied redaction annotations survived.

Marking areas manually and finding patterns automatically

You can combine freehand box drawing with automated pattern matching on the same document. Preview thumbnails render locally at 680 px width so you can inspect every page before committing changes.

Using Find and mark text presets for emails, phones, and numbers

Type any exact name, account ID, or phrase into Find and mark text and select Find to mark every match across all pages. Three built-in preset buttons (Email addresses, Phone numbers, and Long numbers) run regular-expression scans over the page text layer.

If a scanned document has no selectable text layer yet, add a searchable text layer to image scans first or draw boxes manually over the rendered image. When you export, redacted pages render at 200 DPI and encode as JPEG at 0.92 quality using the MDN HTMLCanvasElement toBlob API to keep printed text legible.

Nudging and resizing redaction boxes with the keyboard

Pointer dragging ignores tiny accidental clicks smaller than a fraction of a percent of the page, preventing stray dots. Once a box is placed, click it to focus and use arrow keys to nudge its position, + or - to resize, or Delete to remove it.

Holding the Shift key while pressing arrow keys moves the selection in larger jumps. Select Clear all marks in the options panel if you want to discard all boxes and start fresh.

How PDF redaction scrubs hidden data beyond page text

Deleting visible words from a page is only half of safe document sanitization. PDF files frequently store duplicate copies of sensitive strings inside form fields, accessibility trees, and metadata streams.

PDF structure element Potential data leak Redact PDF behavior
Page content stream Selectable text operators under black boxes Old page removed and replaced with flat 200 DPI image
AcroForm fields and widgets Interactive text box values on redacted pages Pruned from form array and parent field values cleared
Tagged PDF structure tree Alternative and actual text copies in tags Structure tree and mark info dictionaries deleted
Document properties and XMP Author, title, subject, and application history Stripped when Also remove document properties is checked
Outline bookmarks and labels Sidebar chapter names repeating redacted terms Deleted when checked, or matching terms scrubbed

Tagged accessibility structures defined in the Library of Congress PDF 2.0 Format Description can store full text duplicates inside structure elements. Deleting those dictionaries guarantees that screen-reader fallback trees cannot expose words you removed from the page.

If you also want to inspect a document for hidden attachments, annotations, or embedded metadata without blacking out page text, use full PDF privacy and metadata inspection. Combining both checks covers both visual content and container metadata.

Verifying and distributing redacted PDF files safely

Because every rendering and encoding step runs inside your browser tab, confidential medical records, legal discovery files, and tax IDs never leave your device. Still, reviewing your output before external distribution is good operational hygiene.

When to redact versus when to crop page margins

Cropping a PDF changes the visible CropBox boundary while leaving off-screen graphics and text inside the file stream. Never rely on margin cropping to hide confidential headers, footers, or account numbers.

Use Redact PDF whenever sensitive information must be destroyed permanently. When you only need to trim outer whitespace without altering sensitive content, margin cropping preserves vector sharpness across all pages.

Tip: Always test your exported PDF by opening it in a viewer, pressing Ctrl+F or Cmd+F to search for a redacted term, and confirming zero matches appear.

Complete sanitization aligns with the purge principles in the NIST SP 800-88 Rev. 1 Media Sanitization Guidelines, which require removing data from all reachable structures rather than masking display output. Always confirm court, agency, or organizational redaction standards before filing regulated records.

When a document contains multiple pages of sensitive tables or signatures, step through each page with the Previous page and Next page controls before selecting Redact. Keeping untouched pages vectorized while rasterizing only redacted pages delivers both security and compact file sizes.

Questions

Can redacted text be recovered from a flattened PDF image?

No. Each page containing a redaction box is rendered to a flat raster image with solid black pixels burned over the masked region, and the original page object is deleted from the file. Unreachable text streams are pruned before saving.

Does redacting one page affect text selection on other pages?

No. Only pages where you placed at least one redaction box are converted into flat 200 DPI images. Every untouched page stays in its original vector PDF format with selectable text and crisp fonts.

How do I redact emails and phone numbers automatically in a PDF?

Type a word into Find and mark text and select Find, or click Email addresses, Phone numbers, or Long numbers below the search bar. The tool scans every page locally and places editable black boxes over each match.

Why should I remove document properties and bookmarks after redacting?

PDF metadata fields such as Title, Author, and XMP streams, along with sidebar bookmarks, often repeat names or case numbers you blacked out on the page. Keeping both cleanup boxes checked removes those secondary leaks.

Can I use keyboard shortcuts to adjust redaction boxes?

Yes. Click a marked box to focus it, then press arrow keys to nudge its position, hold Shift with arrow keys for larger steps, press plus or minus to resize, or press Delete to remove the mark.

Does this redaction tool work on scanned PDFs without selectable text?

Yes, for manual drawing. You can drag redaction boxes directly over any scanned signature, photo, or printed line. Automatic text search and pattern presets require an embedded text layer on the page.

Updated