Nyxiontech PDF Tools › Redact PDF

How to check whether a PDF tool really uploads your files

Almost every online PDF tool promises something about your privacy. "Processed in your browser", "we delete your files after an hour", "bank-level encryption", "GDPR compliant". Those are four different claims, and only one of them can be checked by the person reading it. This guide shows how to check it, on any tool, in any modern browser, without installing anything.

We build PDF tools ourselves, so read this as an invitation rather than a sales pitch: run these checks on us too. The last section says exactly what you should see when you do.

The check that settles it: watch the network

A file can only leave your computer as a network request, and your browser lists every request it makes. So you do not have to trust anyone's description of what a page does. You can look.

  1. Open the tool's page, but do not choose your file yet.
  2. Open the developer tools: F12, or Ctrl + Shift + I on Windows and Linux, or Cmd + Option + I on a Mac.
  3. Go to the Network panel (Chrome, Firefox) and clear the list.
  4. Now do the real job: choose your file and run the conversion.
  5. Look at what appeared, and sort the list by size.

If your file was uploaded, there is a request carrying it, and it is hard to disguise. It will be a POST or a PUT, its size will be close to the size of your file, and its request payload will be the file itself. A tool that works locally produces nothing of the sort: the page, the scripts and the fonts load at the start, and then the list stays still while your 12 MB scan is converted.

Size is the part to watch. A local tool may still send small requests for analytics or error reporting, and those are not your document. One request roughly the size of your file is the signal. A handful of requests a few kilobytes each is not.

The stronger version: pull the plug

There is a blunter test that needs no reading of request payloads at all. Load the page, disconnect from the internet, then do the job.

  1. Open the tool and let the page finish loading.
  2. Turn off your Wi-Fi, or tick Offline in the Network panel of the developer tools.
  3. Choose your file and run the conversion.

If it still works, your file never needed a server, because at that moment there was no server to reach. If it fails or hangs, the work was happening somewhere else. This is the one check with no room for interpretation, and it takes about fifteen seconds.

Read the policy the browser enforces

A site can declare where its own pages are allowed to send data, in a Content-Security-Policy response header. This is not a sentence in a privacy policy: the browser enforces it and blocks anything outside it. To see it, open the Network panel, click the first entry (the document itself) and look under Response Headers. On the command line, curl -I https://example.com prints the same thing.

Two directives matter here. connect-src lists where the page's scripts may send data (MDN), and form-action lists where a form may submit (MDN). A policy of connect-src 'self' means scripts can only call back to the same site, which rules out the page handing your file to somebody else.

One trap is worth knowing, and MDN states it plainly for form-action: it does not fall back to default-src, so "Not setting this allows anything". A missing directive is not a restriction. Read what is in the header, not what you expected to find there.

Be clear about what this proves. A content security policy constrains the page running in your browser. It says nothing about what a server does with a file you did send it. It is useful evidence next to the network check, and no substitute for it.

Three claims that sound like privacy and are not

Only one claim removes the question: the file never left your computer. That is also the only one you can confirm for yourself, in a minute, with the checks above.

When this matters more than usual

For a holiday photo it hardly matters. It matters when the document belongs to somebody else: a client file, a medical record, an employee contract, a CV sent to you in confidence. Uploading one of those to a third party is not a technical detail, it is a decision to let another company process personal data, with everything that follows from it. If you cannot tell where the file goes, that alone is reason to do the job locally or in an installed program.

Now try it on us

Run all three checks on this site. Open Redact PDF, or any other tool here, and watch the Network panel while you work: the page and its scripts load, and then your file produces no upload. Go offline first and the tools still run, because the code is already in your browser and the document never needs to travel.

The header on every page here is default-src 'self', connect-src 'self' data: blob: and form-action 'none'. The data: and blob: entries deserve explaining rather than glossing over, because they are not servers. A blob: URL points at something held in your browser's own memory (MDN), which is how a finished file is handed to you as a download. No other site appears in that list, and form-action 'none' means no form on the page can submit anywhere at all.

That is the whole point of writing this down. You should not have to believe a sentence on a landing page, ours included, when the browser you already have will show you the answer.

Open Redact PDF