See what a file still says about you before you send it.
Your file is opened here and never uploaded. Asking what a finding means is a separate button, and it sends the recovered text only.
PDF, JPG, PNG, or Word, Excel and PowerPoint. Nothing is uploaded — the report is built here.
Taking the file apart…
Nothing was recovered by the checks below. That is not the same as the file being clean — it is the answer to the questions this tool knows how to ask.
The file is read in this browser tab and is never uploaded. Only if you ask for the plain-English explanation is the recovered text sent — and only that text. The recovered text is used to answer that one request and is not stored. The file itself never leaves your device, so there is nothing of it to keep.
Before a file goes to a landlord, a court, an insurer or a recruiter, it is worth knowing what the person at the other end could recover from it. Not what it looks like — what is in it. Drop a file here and this reports everything it can still give away.
The things people are caught by are always the same. A black rectangle drawn over a name in a PDF editor is a drawing: the name is still underneath, and it copies and pastes straight out. A cropped photo keeps a small copy of the uncropped picture inside it, because cameras write a thumbnail when the shutter fires and most editors never rewrite it. A Word document with tracked changes turned off still contains every deleted sentence. A file's own properties usually name the person who wrote it and the folder it was saved in, which is often somebody's account name and the title of an earlier draft.
Finding all of that happens on your device. Nothing is uploaded, and the report appears without anything crossing the network. If you then want the plain-English reading of it — which of these actually matters, and why — that is a separate button, and it sends the recovered text and nothing else.
There is no version of this that says your file is clean. It tells you what it checked, what it found, and what it could not look at, because the list of ways a document can betray its sender is not a list anyone has finished writing.
The failure that keeps reaching the news — a court filing, a government report, a legal settlement, published with black boxes over the names and the names still in the file. If the text comes back here, the box is decoration.
Cropping removes it from the picture, not from the file. The embedded preview is usually the version from before you touched it, and this shows you both side by side.
Contracts, proposals and reports get reused. Tracked deletions, comments, and the previous client's name in the properties all travel with the file, and none of them are visible on the page.
If location services were on, the coordinates of where you were standing are in the file, to six decimal places. That is the front door on a map.
Form fields keep their values as data, separate from what is drawn on the page. A field you cleared on screen can still be sitting in the file.
A PDF, a photo, or a Word, Excel or PowerPoint file. It is read in this browser tab — nothing is uploaded, and the report appears without the file going anywhere.
Each row is something recoverable from the file, with where it was found and the text itself. Underneath, the checks that were run and the things this could not look at.
One button sends the recovered text — not the file — to be sorted by how much it would matter if the recipient read it. It is optional, and the findings stand without it.
Pages with hidden text are flattened to pixels, metadata is cleared, annotations are dropped, and photos are rewritten without their metadata segments. What was changed is listed exactly, along with anything deliberately left alone.
No. Everything in the findings list is recovered in this browser tab, and the clean copy is written here too. The one thing that can leave your device is the recovered text, if you press the button asking what it means — and that is text the tool has already shown you, never the file.
It can read anything that was never actually removed, in any file you open with it. That is the point: a redaction that software can undo was not a redaction, and the way to find out is to check. Use it on what you are about to send.
It means every check listed on the page came back empty, and the page names them so you can see which ones they were. It does not mean the file is clean, because no tool can tell you that — new ways to leave something behind in a document turn up regularly, and this knows about the ones it knows about.
Your camera wrote a small preview into the file when it took the picture. Cropping or editing changes the picture; most software does not rewrite the preview. So the file can carry a copy of what the photograph looked like before you changed it, and that copy is what some websites and messaging apps display.
Because that is the only way to be sure. Removing text from a region of a PDF means rewriting the content stream around it, and getting that slightly wrong produces a file that looks redacted and is not — the exact problem you came here about. Turning those pages into images cannot half-work. Only pages with something hidden on them are flattened; the rest keep their text.
A tracked deletion is text somebody already decided to cut, so removing it changes nothing anyone can see. A tracked insertion is text that is in the document everybody reads, so it is unwrapped and kept. Deleting it would change what the document says, which is not this tool's decision to make.
No. If a page is a picture of a document, anything on it is part of the picture and is not recovered here. Image to Text will read it if you need to know what it says.