HowtoredactaPDFproperly

Why the black box fails

A PDF holds content in layers of instructions. Drawing a filled rectangle adds an instruction to paint black pixels; it does not delete the text underneath, which remains in the file exactly as it was.

Anyone can select the region and copy the text out, extract the text layer, or open the file in an editor and move the rectangle. No skill is required, which is why this keeps happening publicly.

The other ways redactions leak

• Metadata — the author, the original filename, the software that produced it.
• Annotations and comments, which are stored separately from the page and are easy to forget.
• Cropping, which hides content by changing the visible box while leaving it in the file.
• Earlier versions inside the same file, where incremental saves have preserved what was there before.

Two methods that actually work

The first: remove the page entirely. If the sensitive content is confined to particular pages, deleting them removes it from the file rather than covering it.

The second: cover, then flatten. Draw over what must go, then flatten the document so every page is re-rendered as an image. The text underneath is not hidden — it no longer exists in the output, because the output is a picture of the page as it now appears.

Flattening costs you the text layer for the whole document, which is the price of certainty. For anything genuinely sensitive, it is the right price.

Then verify it

• Open the redacted file and try to select the covered text. If anything selects, you have not redacted it.
• Extract the text layer and search it for what should be gone.
• Check the metadata, and strip it.
• Compare the page count against the original.

Verification takes a minute. Discovering a failed redaction after distribution cannot be undone — the document is already out.