Where hidden data hides in a PDF
A PDF is a container, and the visible page is only part of what it carries. Documents have leaked from every one of the places below while the page itself was perfectly redacted, which is why removing text from the page is necessary but not the end of the job.
Document properties and XMP
Every PDF has a properties block — title, author, subject, keywords, producer, creation and modification dates — filled in by whatever program made it. The author field commonly carries a real person's name or a network username. The title frequently carries the original filename, which in a professional setting often carries a client name or a matter number. Alongside it many files carry an XMP packet, a block of structured metadata that can hold a longer history including previous titles and authoring tools. Neither block is visible on the page and both travel with the file.
Annotations, comments and form fields
Comments, sticky notes, highlights and review markup are stored separately from the page content and are trivially readable. Form fields are worse: a filled field keeps its value as data, so a form that looks blank in a flattened-looking file can still carry what was typed into it. And an annotation that has been visually removed is sometimes only marked as hidden rather than deleted.
Attachments, layers and things left off-screen
PDFs can carry embedded files — a spreadsheet attached to a report is a common one — which are not shown on any page and are extracted with one click. Optional content groups, which most people meet as layers, can hold content that is switched off for display and still fully present in the file. Content can also sit outside the visible page area, cropped from view rather than removed: cropping changes the page box, it does not delete what falls outside it, and undoing a crop is a menu item.
Incremental saves and previous versions
PDFs support saving changes by appending to the file rather than rewriting it. That means a document that has been edited and re-saved several times can carry its earlier states inside the same file, recoverable by anyone who knows to look. This is one of the reasons this app never edits the loaded document and saves it back over itself. The output is always assembled as a new document with pages copied or rebuilt into it, so there is no earlier state to append to and no orphaned objects from the original left behind.
Frequently asked questions
Where is hidden text stored in a PDF?
Most often in the document properties, an XMP metadata packet, annotations and comments, form field values, embedded attachments, hidden layers, content cropped out of view, or earlier revisions kept by incremental saving.
Does removing text from the page clear the metadata?
No. They are separate parts of the file. Metadata has to be cleared on its own.
Can a cropped area be recovered?
Yes. Cropping changes which part of the page is displayed; it does not delete the content outside it, and the crop can be undone.
What about a filled-in form that looks empty?
Field values are stored as data. A field can display as empty and still carry what was typed into it.
Do embedded attachments show up when I read the file?
Not on the page. They are listed in the viewer's attachments panel and extracted with a click.
How do I know whether my file has any of this?
Read the document properties in any viewer, check the attachments and layers panels, and run a raw scan of the file's streams for the terms you care about.
Remove the text, then prove it is gone. Redact takes the words out of the page rather than covering them, then reopens the file it just made and searches it twice — by text extraction and by raw stream scan — before letting you download. The file never leaves your browser.
Redact a PDF in the browser