Updated on October 8, 2026
How a PDF stores its text
A PDF isn't a word-processing document. Each page is a list of drawing instructions: “move here, use this font, draw these characters”, “draw this rectangle”, “show this image”. The text you see is the result of those instructions, drawn one after another, in order.
The direct consequence: whatever is drawn last covers whatever was drawn before. A white rectangle placed after a line of text hides it on screen… without removing it from the list of instructions.
Method 1: the overlay, or the white box
This is how most online PDF editors and annotations work: a box the color of the background is added over the old text, then the new text on top of the box. Nothing is removed; everything is added.
So the file holds three stacked layers: the old text, the box, the new text. And each of them stays usable:
- search and copy-paste return the old text, sometimes mixed with the new one;
- screen readers (accessibility) often read both versions;
- any PDF editor can move or delete the box and reveal the old text;
- on a colored background, a photo or a table, the box shows.
It's exactly the mechanism behind the failed redactions that regularly make the news: documents published with passages “blacked out” by a black rectangle, whose text can be recovered with a simple copy-paste. In January 2019, a court filing by Paul Manafort's lawyers revealed information that was supposed to be hidden this way (source).
Method 2: real editing
Really editing a PDF means removing the instructions that draw the old text from the page, then adding the ones that draw the new text. Afterwards, the old text no longer exists in the file: there is nothing to reveal, and a search returns the new value.
It's harder than it sounds, which is why few tools do it:
- fonts are often incomplete: a PDF usually embeds only the letters it uses. Writing a new word sometimes needs a replacement font with the same letter widths, or the line overflows or shifts;
- a line can mix several styles: a bold word in the middle of a sentence must stay bold when the rest is edited;
- nothing may be erased next door: in a tight table, the areas of neighbouring lines often overlap by a few points;
- some PDFs don't let their text be read back: the characters displayed don't match the ones stored, and the line then has to be retyped.
Glyph uses this method: it removes the original characters (with a “redaction” operation, which actually deletes the content under an area), then reinserts the new text at the exact position of the old one, in the original font or the closest one. The final file is entirely rewritten: the old text doesn't survive in its internal data either.
How to check a PDF yourself
- Open the PDF, select the text around the edited or blacked-out area, and paste it into a text editor. If the old text shows up, it's still in the file.
- Use search (Ctrl+F or Cmd+F) with a word from the old text.
- Open the file in a PDF editor and try moving the boxes: an overlay moves, a real deletion leaves nothing behind.
- Look at the document properties: the title, author or keywords sometimes hold information you thought was removed.
A PDF can also keep older versions of itself when it was edited “by appending” (incremental save). Removing them requires rewriting the whole file, which Glyph does. To check everything at once, the redacted PDF checker analyses these points for you.
In short
- an overlay hides: the old text stays in the file, recoverable by anyone;
- real editing replaces: the old text disappears from the file;
- to redact a document before sharing it, only real deletion actually protects the information.
To edit a PDF this way: edit the text in a PDF. To hide information for good: redact a PDF.