Each finding comes with its page, a snippet masked to the first and last character of each word, and a severity: high when the hidden text contains an email address, a Social Security, card or IBAN number, or a date of birth written next to a label. Fix in Redact opens Redact with a box over every finding that has a place on a page, and those pages are re-rendered without the text. Sanitize removes the rest — earlier versions, orphaned objects, hidden field values, thumbnails, metadata, attachments and scripts — and checks the file again. Two things can look hidden without being a leak: the invisible layer OCR lays over a scan, which is not reported where the scan shows the printed words under it, and text printed very faint on purpose, such as a pale watermark, which can be reported as covered, so look at the page before you box it. Checking uses no job, and the file never leaves your device.
How it works
- Drop the PDF and open Leak Check — every page is rendered and measured on your device
- Read the findings: covered, tiny and off-page text, hidden fields and layers, earlier versions and orphaned objects, each with its page
- Box them with Fix in Redact, remove the rest with Sanitize, and check the result again
Questions people ask
Where can text hide in a PDF?
Under something: a box, a picture or a comment drawn over it, or white text on a white fill. Out of sight: text under 1 point high, or placed beyond the edge of the page. In the file’s structure: form fields that are hidden or set not to display, layers switched off in the default view, page thumbnails, metadata, attachments, earlier versions kept by incremental saves, and objects from pages that were replaced or deleted but never cleared out. Leak Check looks in each of these, and lists redaction marks that were made but never applied.
Will the OCR text layer in a scanned PDF be reported?
OCR puts invisible words over the scan on purpose, so the scan can be searched. Leak Check expects that: where the rendered page shows the printed words under the layer, those words are not reported as covered. A scan with no text layer at all has nothing for the text checks to read — its words are pixels — so only the checks that read the file’s structure apply to it.
Can it find words hidden inside a picture?
No. Leak Check reads text, not images. Words that are part of a photo or a scanned page are pixels, and a box drawn over part of a picture leaves the covered pixels in the picture underneath. Redact removes those, because a redacted page is re-rendered as one flat image. Text a picture covers only in part can also be missed: the uncovered part still shows ink, so the word counts as visible.