Covering text with a black rectangle is not redaction. The words are still in the file, selectable and searchable. Here is how to actually remove sensitive information from a PDF.
Redaction is one of the most misunderstood operations in document handling. The classic mistake — drawing a black box over the text you want to hide — produces a file that looks redacted but is not. The underlying text, images, and metadata are still present, fully searchable, and trivially extractable. Real redaction means deleting the content from the file structure.
Why black boxes do not work
A PDF is a structured collection of objects: text runs, vector paths, raster images, and metadata. A black rectangle drawn on top is just another object layered above the text. Select the area with a text tool, copy, and paste into a text editor, and the “hidden” words appear. Export the PDF to plain text, and the redacted content is right there. The visual layer and the data layer are separate, and covering one does not touch the other.
What true redaction does
Genuine redaction removes the text objects themselves from the content stream and replaces them with nothing (or with a marker like “[REDACTED]”). The characters are gone from the file, so there is nothing to select, search, or extract. If the sensitive content is an image — a face, a signature, a license plate — the image region is removed or permanently obscured by re-encoding the pixels, not by overlaying a shape.
A practical redaction workflow
Step 1: Identify every instance
Search the document for the term you need to remove. A name might appear in the body text, in a header, in a footnote, and in metadata. Search catches the obvious occurrences; metadata is easy to miss. Check the document properties for the author, title, and keywords fields, which often carry names and internal project codes.
Step 2: Remove the content at the source
If you have the source document (a Word file, a design file), redact there and re-export to PDF. This is the cleanest path because the content never makes it into the PDF. When you only have the PDF, use a redaction tool that deletes the underlying text objects, not one that paints over them.
Step 3: Flatten and re-export
After removing the text, re-export or “flatten” the PDF so the content streams are rewritten without the deleted objects. This is also the moment to strip metadata. Re-running the file through PDF Compressor or reassembling it with Merge PDF produces a fresh file with a clean object structure and no leftover references to the removed content.
Step 4: Verify
Open the result, try to select the redacted regions, search for the original terms, and export to plain text. If any of those reveal the content, the redaction failed. Repeat until the file is clean.
Common redaction failures
- Black rectangles layered over intact text — the most common and most dangerous failure.
- Redacted body text but unredacted metadata (author, title, comments).
- Redacted text but a copy of the page kept as an image layer underneath.
- Redaction applied to one page but the same content repeated in an appendix.
- A “redacted” file that still contains the original as an embedded attachment.
Redaction and OCR
Scanned PDFs are images, not text, so searching and selecting do not work — but OCR can reconstruct a text layer from the pixels. If you redact a scan by drawing a box, run the file through Image to Text OCR afterward to confirm the obscured words are not recoverable. If the OCR reads them, the redaction is visual only and the pixels must be permanently removed or re-encoded.
Redaction is a delete operation, not a paint operation. Treat it that way, verify the result, and never trust a black box.