Redacting a document so it stays redacted
Redaction failures are among the most reliably embarrassing incidents there are, because the document is usually already public by the time anybody notices and it cannot be recalled.
- 01
Redact the content, do not cover it
minutesUse a tool with a genuine redaction function that removes the underlying characters rather than an annotation that draws over them. In Acrobat this is Redact and then Apply, and the apply step is the one people miss. Anything that leaves a shape you can select and delete has not redacted anything.
how to know it workedOpen the finished file and try to select the text under the mark, then run it through a text extractor. If any of the removed words come back, it is not redacted.
- 02
Strip the metadata, which is a separate step
minutesAuthor, organisation, revision history, tracked changes, comments, the original filename and the path it was saved from. Office documents routinely carry earlier versions of the text that was edited out. None of this is removed by redacting the visible body.
how to know it workedexiftool on the finished file, or Document Inspector in Office. Read the output rather than trusting that the tool ran.
- 03
Flatten, or print to a new file
minutesPrinting to PDF or exporting flattened destroys layers, annotations and anything hiding behind an image. It also destroys accessibility structure, so it is a trade rather than a free win: for a document that must stay screen-reader navigable, use real redaction and verify it instead.
how to know it workedCompare the file size before and after. A flattened document that is the same size as the original probably kept what you thought you removed.
- 04
Watch for the things that are not text
an afternoonA chart whose underlying spreadsheet is embedded. A photograph with GPS coordinates. A screenshot with a different window visible at the edge, or a hostname in a title bar. A cropped image that is cropped by display rather than by data, so the removed part is still in the file.
how to know it workedExtract every embedded object and every image and look at them individually. The failure is almost never in the paragraph somebody redacted carefully.
- 05
Have somebody else check it
minutesThe person who redacted it knows what is meant to be gone and reads past it. A second person, told only "tell me anything you can recover", finds what the first missed. For anything going outside the organisation this should be a rule rather than a courtesy.
how to know it workedThe check happened and is recorded. On a disclosure that matters, who checked it is a question that gets asked afterwards.
what this does not cover
- Inference. Redacting a name while leaving the job title, the office and the date can identify somebody as precisely as the name did
- The copy that was already sent, in a mailbox you do not control
- The version in a document management system, which usually keeps every revision including the one before the redaction