Explainer
Why redaction fails
In 2025, when congressional releases of documents from the Jeffrey Epstein estate went out with black boxes over the sensitive parts, readers found within hours — and news outlets confirmed — that some of the boxes were only drawn on top of the text. Selecting the “redacted” passages and copying them pasted the hidden names and details into any text editor.
The mistake was not new. In 2019, lawyers for Paul Manafort filed a court document with passages blacked out; journalists selected the bars, copied the text beneath them, and published what it said — including material about the sharing of polling data — because the redactions were drawn over live text rather than applied to it.
Both files looked redacted. Both were still readable. And both failures were made by professionals with lawyers in the room. Here is why it keeps happening — and how to check your own files before someone else does.
The page is not the file
A PDF is not a picture of a page; it is a program that draws one. Text, images, and that black rectangle are separate instructions, and drawing a box after the text changes what the page looks like, not what the file contains. Office documents are further still from their appearance: a .docx, .xlsx, or .pptx is a zip archive of XML parts, and much of what those parts remember — deleted text, hidden sheets, the author of every comment — is never painted on screen at all. Screens show the page. Files carry the history.
The eight ways a “redacted” file leaks
- A box drawn over live text
The classic. The black rectangle is one more drawing instruction, painted after the text — the text object underneath survives, and select-all, copy, or any extraction tool reads straight through the ink. The same goes for white-on-white text and text shrunk below visibility: styled, not removed.
- Redaction marked, never applied
Real PDF redaction is a two-step tool: mark the areas, then apply. The mark is an annotation that says “redact this” while the content sits untouched beneath it. Skip the second step and the file ships with a labelled map to its own secrets.
- Metadata
Author names, company, the template path on a shared drive, total editing time, creation and revision dates, the exact software used — carried in the PDF Info dictionary, XMP, and Office document properties, and shipped with every copy.
- Comments and annotations
Review threads travel with the file: who commented, when, and what they actually said about the matter — often franker than anything in the body text.
- Attachments and embedded objects
A PDF or Office file can carry whole other files inside it — the source spreadsheet embedded in the report, complete with the rows the report left out.
- Prior versions
PDFs can be saved incrementally: the new version is appended, not substituted. The “deleted” earlier document is still in the byte stream and can be recovered by reading the file as it stood one save ago.
- Hidden layers and hidden content
PDF layers switched off, Excel sheets marked hidden or “very hidden”, hidden rows and slides, speaker notes, text formatted as invisible. Out of sight is not out of file.
- Cropped images and thumbnails
A crop in Office is a viewport, not a cut — the full original picture ships in the package. Photos carry an embedded preview thumbnail that can predate the edit you made to the visible image.
How to check
Open the file, and look for what a reader could still get out. Manually, that means extracting text under every dark region, unzipping the Office package and reading its parts, and walking the metadata — for every file, every time. The scanner on our front page does all of it in one drop: it reads the file inside your browser — nothing is uploaded, and you can watch the network panel to confirm — and shows everything still recoverable, including the recovered content itself: the text under the boxes, the deleted paragraphs, the hidden sheet’s rows, the coordinates in the photo. The check is free and unlimited.
How to fix it
The only redaction that works is the one that destroys the content. For PDFs the dependable fix is flattening: render each page to plain pixels and rebuild the document from those images, so nothing under the boxes survives — because nothing under the boxes is kept. For Office files, the fix is removing the leaking parts — accept the tracked changes, delete the comments, the authors, the hidden sheets, the properties — and rewriting the package. For photos, re-encoding the pixels strips the GPS, the serial numbers, and the embedded thumbnail. StillReadable Pro does all of this in your browser too, then re-scans the cleaned file and shows you before and after, so you are not taking anyone’s word for it — including ours.
Drop in the file you were about to send. Nothing is uploaded.
Check a file — freeStillReadable finds what is still recoverable; it is not legal advice or a compliance guarantee, and no automated check replaces professional review of a sensitive release.