Is It Safe to Upload PDFs to Online Tools?
Uploading a document to a website means it will exist somewhere that is not your computer — even if only for a moment. For a seating chart, who cares. For a passport scan, a signed contract or medical records, the questions below are the difference between convenience and a problem.
The five questions that matter
- How long is the file kept? “Forever, in your account” and “deleted automatically” are opposite answers. Look for a stated retention window in the privacy policy, not just the FAQ.
- Can anyone look inside? Automated processing is normal; human review is not. A service that says nothing about human access has not told you it doesn't happen.
- What is done with the content? Training machine learning models on uploaded documents is a real practice now. It should be stated clearly — in either direction.
- Who else touches it? Hosting, CDNs and email providers are normal processors; the privacy policy should list the categories and the reasons.
- How is it protected in transit and at rest? HTTPS is the floor. Anything sensitive deserves a service that says plainly where files live and for how long.
How our own tools answer
We built the answers into the product, and they are written down in the Privacy Policy:
- Files are stored under an unguessable id and deleted automatically once they are no longer in use — nothing is archived.
- No human review, ever: nobody on the team opens documents; the only access is automated processing.
- No training of anything on your documents.
- OCR runs in your browser — a scanned page isn't even uploaded for recognition.
- No ad trackers, no third-party analytics profiling, no data selling.
The Security page lists the same controls in more detail, including retention, hashing and how to report a vulnerability.
When to go further than any website
Some documents deserve not leaving the machine at all: unredacted IDs, medical records, anything under NDA. Two solid options: redact before you upload (a whiteout in the editor genuinely covers the content region in the output), or run the toolkit yourself — the whole project is self-hostable, which is the only answer that involves no third party at all.
The one-minute test
Before using any unfamiliar PDF site, spend a minute on its privacy page looking for the retention window and the human-access sentence. If both are missing, that is your answer — and there is usually a neighbour in the search results that says both plainly.
Questions to ask any PDF site
- Is the connection HTTPS, and is the upload encrypted end to end?
- How long is the file kept, and does opening it reset the clock?
- Can a human browse or open uploaded files?
- Is processing automated, or is the document sent to another provider?
- Does the retention policy apply to backups and logs as well?
If the answers are vague, the site is asking for trust without evidence. A clear retention window and a plain statement that files are not browsed are the minimum.
What “deleted” should mean
Deletion after a fixed window is better than “kept indefinitely”, but the detail matters: is it hours of no use, or hours since upload? Is the clock reset by an open tab? A good policy explains the rule and applies it to the working copy. A PDF editor is not a document archive; export the result and keep your own master.
Red flags
Requests for an account before you can see a price, unclear ownership, a download that is an executable, or a tool that cannot explain where processing happens are all reasons to leave. For a sensitive document, prefer a workflow that keeps recognition in the browser, or one with a named retention window and a security page. The Security page explains the controls used here.