How to Check Whether a PDF Tool Uploads Your Files
To check whether a PDF tool uploads files, open your browser's DevTools Network panel, clear its requests, run the tool with a test PDF, and inspect any POST, PUT, or multipart requests whose size resembles the file. Then turn off the network after the page loads and try again. Read the provider's privacy policy for upload and retention terms; a small analytics ping alone is not a file upload.
Start with a harmless test file
Make a small PDF containing only dummy text, such as a sample receipt or two-page note. Record its approximate file size so you can recognize a transfer later. Do not use a confidential document for an experiment with an unfamiliar website; a successful test might itself send the file. Open the tool page and let its scripts and fonts finish loading before you begin measuring the processing step.
A browser can contact a site for ordinary reasons while a page loads. The aim is to observe what happens when you choose and process the PDF, not to declare every network request suspicious. Use a separate test file for each feature you want to check: merging, compression, OCR, and cloud AI may have different data paths even on the same website.
Inspect the Network panel
Open DevTools with your browser's menu or keyboard shortcut, select Network, and clear the existing request list. Keep the panel open while you add the test PDF and run the action. Look especially for POST or PUT requests, multipart/form-data uploads, or request bodies that include a file field. Compare the request's transferred or payload size with the PDF's size; a large request around the moment you process the file deserves closer inspection.
One request can carry a file in a format other than multipart, and some tools split work into multiple requests. Inspect the request details and timing rather than relying on a single label. A tiny analytics or error-reporting ping is not itself evidence that the PDF was uploaded, although it may still matter to your privacy preferences. A large response might simply be a downloaded output; focus on outgoing request bodies and where they go.
Try the tool offline
Load the tool page first, then disconnect the network or select the browser's Offline mode. Clear the Network panel and repeat the task with the harmless PDF. If processing and download still work, that is useful evidence that this particular action can run locally with the resources already loaded. If it fails, the failure may be a missing script, font, worker, license check, or server processing, so investigate before drawing a firm conclusion.
Repeat the test with any special feature you plan to use. A site's ordinary merge operation may run locally while OCR or an AI summary needs a service. Cached pages are important: a tool cannot demonstrate offline processing if its code was never downloaded. Very large PDFs may also fail locally because browser memory is limited, even when the tool does not upload them.
Read the policy and feature notices
Find the provider's privacy or security page and search for upload, storage, deletion, retention, and account-save terms. Look for concrete windows, such as deletion after processing, and for exceptions involving signatures or saved files. A statement that a tool 'runs in your browser' can refer only to its interface; it does not prove the PDF stays there. The Network test and the written policy answer different parts of the question, so use both.
If the policy is unclear, ask the provider which feature receives file bytes and for how long they remain available. Also check the destination of any upload: a third-party processor may be involved. Policies can change, so repeat this review when your organization's rules require it or when a provider changes its workflow. Keep a dated note of what you observed if you need to explain the decision to a colleague.
A short worked example
Paperlane's core Merge PDF tool processes file bytes in the browser without uploading them. Load the page, clear the Network panel, add two harmless PDFs, merge, and check outgoing requests while the result is made. Then turn on Offline mode and repeat after the page has been cached. The merge should still work. Paperlane is free, requires no account, and adds no watermark to that result.
This observation applies to the core local tool you tested. Paperlane's peer-to-peer sharing is the one feature that opens a connection between browsers, so test it separately if you choose to use it. Paperlane does not offer password encryption or unlocking. For any provider, the safest habit is to verify the specific action with dummy data before using a file that matters.
Frequently asked questions
Does a POST request always mean my PDF was uploaded?
No. Check its destination, timing, and payload size. A small analytics request can be unrelated to file bytes, while a large POST during processing needs closer inspection.
Is an offline test enough by itself?
It is useful evidence for the feature you tried after loading the page, but it does not describe every feature or the provider's retention policy. Review Network requests and written terms too.
Why does a local tool fail on a huge PDF?
Browser processing uses your device's memory. A file can exceed available RAM even if no bytes are sent to a server.
Can a site say 'runs in browser' and still upload files?
Yes. The interface may run in the browser while a particular operation sends the PDF for server processing. Test the operation itself with a harmless file.