TabTasker
The journey of an uploaded PDF
You've got a 200-page PDF that needs compressing. You find a tool online, drop the file in, and wait. A moment later, a smaller version downloads. Problem solved, right?
Let's look at what actually happened behind the scenes.
Step 1: Upload
Your PDF travels from your computer, through your internet provider, across multiple network hops, to the tool's server. During this trip, it passes through infrastructure that could theoretically inspect it, though most tools use HTTPS which encrypts the transit.
The real issue starts when the file arrives.
Step 2: Storage (even temporary)
The server needs to store your file somewhere to process it. This might be a temporary directory, an object store, or a database. The tool's documentation might say "files are deleted after processing" or "we keep files for 2 hours."
Here's the thing about temporary storage: temporary isn't the same as nonexistent. During that window, your file is sitting on someone else's infrastructure.
Step 3: Processing
The server runs its compression algorithm. This step is usually fine, but it means the server software has full access to every page, every word, every piece of data in your document.
Step 4: "Deletion"
When the tool says it deletes your file, it usually means it removes the reference to it. But file deletion on servers is complicated:
- Backups. Most servers run automated backups. Your file might exist in a backup for days or weeks.
- Logs. File names, sizes, and metadata often end up in server logs that are retained much longer than the files themselves.
- Disk persistence. Even after a file is "deleted," the data can remain on disk until it's overwritten, and forensic tools can recover it.
Real-world examples
This isn't theoretical. There have been multiple incidents:
- In 2022, a popular online PDF tool was found to be sending user documents to third-party analytics services.
- In 2023, researchers discovered that several "temporary" file storage services were retaining documents indefinitely.
- Multiple file-sharing and conversion services have suffered breaches exposing user uploads.
The alternative: never upload in the first place
Tools like TabTasker's PDF Compressor run entirely in your browser using WebAssembly. Your PDF never leaves your device. There's no server to breach, no backup to leak, no log to expose. The same goes for PDF Merge, PDF Split, PDF Protect, PDF Unlock, PDF Redact, and PDF OCR, all of which process files locally in the tab.
The compression happens in your browser's memory. When you close the tab, it's gone. Not "deleted from a server," but actually gone.
Bottom line
If a document is sensitive enough that you wouldn't email it to a stranger, don't upload it to a web server either. Use tools that keep your files where they belong: on your own device.
The privacy policy problem
Most online PDF tools have privacy policies that say something like "we delete files after processing." But there are several things these policies rarely address:
- Sub-processors. The tool may use third-party cloud storage (AWS S3, Google Cloud) to hold your files. Your data is now on multiple parties' infrastructure.
- Staff access. Engineers troubleshooting a bug might have access to file storage logs or the files themselves.
- Legal holds. If the company receives a legal request, they may be required to preserve data that would otherwise have been deleted.
- Acquisitions. If the tool is acquired, the new owner inherits the data, and may not share the previous operator's commitment to deletion.
What "client-side" actually means for PDFs
When a PDF tool claims to be client-side, it means the processing happens in your browser using WebAssembly, the same approach used by TabTasker's PDF tools. You can verify this yourself: open your browser's Network panel in developer tools (F12), drop in your PDF, and watch. If you see your file being transmitted, it's not truly client-side.
A genuinely local tool will show no file upload in the network tab. The PDF bytes stay in your browser's memory, processed by JavaScript and WASM, and discarded when you close the tab.