What Does Redact Mean?

Redaction began as an editorial task: selecting and arranging material for publication. The modern usage narrowed to the removal step, and it entered legal and governmental vocabulary because institutions needed a formal term for releasing a document with specific parts withheld.

That origin explains a distinction that still matters. Redaction is not the same as censorship. Censorship suppresses material that would otherwise be published in full. Redaction withholds a subset of a document that is otherwise released. In practice the line is blurry, but the technical meaning is narrower: the released artifact is the original minus the removed parts.

The word is also used loosely in everyday speech. 'Redacted' on a social media post often means 'covered with a black bar', which may or may not be true removal. That looseness is the source of most of the confusion this page exists to clear up.

When a court or an agency releases a redacted document, the requirement is that the withheld content cannot be recovered from the released copy. Applying a black rectangle over text in a PDF does not satisfy that requirement, because the text layer remains and can be extracted. There is a documented history of exactly this failure in filings and freedom-of-information releases.

The technical standard that follows is the one worth borrowing for personal use: the released file must not contain the removed content in any recoverable form. Not in a layer, not in a text run, not in an embedded thumbnail, not in a revision history.

This is why 'flatten and re-encode' is the phrase that shows up in document-handling guidance. Flattening collapses layers into pixels; re-encoding writes a new file from those pixels. Both steps are load-bearing.

It is not blurring, which degrades rather than removes. It is not pixelation, which averages. It is not covering with a shape, which leaves everything underneath. It is not cropping, which changes the frame but leaves the container and any embedded copies.

It is also not anonymization in the formal sense. Anonymization means the data can no longer be linked to an individual by any means. Redaction removes specific fields. A photo with the face and the EXIF location removed can still identify a place through background details, and a screenshot with names covered can still identify a company through the interface.

Keeping that distinction clear is useful, because it prevents the common overconfidence after a successful redaction. Removing the field you were worried about is a good step. It is not a guarantee that the artifact is anonymous.

An image has no text layer, so the failure mode of overlay-based PDF redaction does not apply. What does apply is the choice of operation. Filling a region with an opaque color overwrites the pixels; blurring or pixelating transforms them and leaves a residual.

A second consideration specific to images is the export path. If the editor composites boxes in memory and the export flattens that composite, the result is a clean file. If the editor exports the original plus a separate overlay object, the result is not. This tool draws everything into a single canvas and encodes that canvas, so there is no second object in the output.

For documents that arrive as PDFs, rasterize first. Converting the page to an image and then redacting the image is the only route that reliably produces a released file without a readable text layer.