The usual model
Your file → upload → their server → download → you
Your document sits on someone else's disk for as long as their retention policy says. You are trusting a privacy policy, not an architecture.
Every PDF site claims your file is safe. Don't take our word for it: open your browser's Network tab, run a tool, and watch the request list. Your file will never appear in it — because it never leaves your device.
The usual model
Your file → upload → their server → download → you
Your document sits on someone else's disk for as long as their retention policy says. You are trusting a privacy policy, not an architecture.
BytesPDF
Your file → your CPU → your browser tab → you
No upload step exists to trust. The processing happens on the machine in front of you, and the finished file downloads straight back to it.
Works in Chrome, Edge, Firefox and Safari on desktop. Keyboard shortcuts below are for Chrome and Edge; Firefox and Safari use the same panels.
We would rather explain every row in that list than let you wonder about one.
JavaScript, CSS, WebAssembly codecs and web-worker files served from bytespdf.com. These are the tools themselves — they have to download once before the work can happen on your device.
Page views and coarse tool/download events. They carry no PDF bytes, no filename, no file size and no account ID. Decline analytics in the consent banner and they stop entirely.
Account and Pro subscription data. Nothing to do with your documents: there is no upload endpoint for PDF content on BytesPDF.
That is the whole architecture. There is no server-side upload API for document content, so a breach cannot leak files we never received.
Load a tool page once while you are online, then switch off your Wi-Fi (or tick offline mode in DevTools) and compress a PDF. It still completes. A tool that sent your document to a server could not finish the job with no connection — this is the shortest disproof of "upload first, trust us later" we know.
BytesPDF needs a connection to load the site and its code. Once the code is in the tab, your document needs nothing from us.
No. Your PDF is read into this tab's memory and processed there using WebAssembly. BytesPDF has no server-side upload endpoint for document content, so your file cannot reach us. The Network-tab test above shows the proof on your own machine, not on ours.
It stays in this tab's memory until you close or reload the page, then the browser releases it. BytesPDF never stores it, never backs it up and never writes it to browser storage. The copy you download lives wherever you saved it, on your device.
No. Every core tool is free end to end: drop a file, process it, download it. There is no signup wall, no email confirmation and no watermark on the output. BytesPDF Pro — an optional subscription for target-size compression, batch jobs, OCR and signatures — does need an account, because that is how a subscription is linked to you.
Yes. Load a tool page once, then turn off your Wi-Fi (or use your browser's offline mode) and compress a PDF. It still works. A tool that uploaded your file could not complete the job with no connection.
Your device is still part of the security boundary: your operating system, browser, extensions, and what you do with the downloaded file afterwards are all outside our control. This page proves one specific claim — that BytesPDF does not receive your document. See the Security page for the full threat model.
Scepticism is the correct default for a tool that touches your contracts, bank statements and medical records. Keep testing us: run the Network-tab check on any page, any time. If you ever see a request carrying your document, email security@bytespdf.com and we will publish the finding.