How to sanitize a file before you share it

The three things that leak: metadata, tracked changes, and content that was covered rather than removed.

Accidental disclosure is rarely dramatic. It is a document sent with the author's name attached, a spreadsheet with a hidden column, or a PDF where a black rectangle sits on top of legible text.

The checklist

Metadata. Nearly every format carries it — author, software, timestamps, sometimes the original filename or GPS coordinates. Check and strip it.

Revision history and comments. Word documents carry tracked changes and comments that do not appear when printed. Accept or reject them, and delete comments, before sharing.

Hidden content. Hidden rows, columns and sheets in spreadsheets. White text on white. Cropped images that still contain the cropped-away part.

Anything you covered up. A rectangle drawn over text hides it from your eyes only.

The mistake worth dwelling on

Covering something with a black box in a PDF leaves the text in the file. Anyone can select it, copy it, or extract it with a script. This has produced court filings, government reports and legal settlements being read in full by journalists within hours of publication, repeatedly, for twenty years.

Cropped images

Cropping in many editors hides part of the picture rather than deleting it, and the original can be recovered by pulling the image out of the document. If a crop is removing something sensitive, export a flattened copy rather than relying on the crop.

Verify, do not assume

Open the finished file the way a recipient would, and look. Try selecting text over redactions, unhiding rows, and inspecting the metadata. Two minutes of checking is worth considerably more than the assumption that a tool did what you intended.

Frequently asked questions

Is removing metadata enough?

No. Metadata is one of three problems. Tracked changes and content that was covered rather than deleted are just as common and usually more damaging.

How do I check a redaction worked?

Open the file and try to select text over the redacted area. If anything selects or copies, the content is still there.

Do images in documents carry location data?

They can. A photograph taken on a phone often holds GPS coordinates, and inserting it into a document may carry that with it.