FileTransmuteFree online file converter
Sign up

Security & privacy architecture

The honest, technical version of how File Transmute handles your files — including where we do use a server, what we collect, and what our tools cannot do.

Last reviewed: July 2026

In-browser tools never upload your file

Most conversions run entirely inside your browser tab. Your file is read into memory with the File API, transformed by JavaScript or a WebAssembly-compiled engine, and handed back as a download. No HTTP request ever carries it anywhere — so privacy is a property of the architecture, not a promise in a policy.

The engines that make that possible

Images use the browser's Canvas API. PDFs are read with pdf.js and written with pdf-lib. Audio and video run through FFmpeg compiled to WebAssembly. OCR, text-to-speech, and background removal run small neural models locally, with WebGPU acceleration where available.

The few tools that do use a server — and why

A small number of conversions genuinely cannot run in a browser: high-fidelity Office rendering needs a real LibreOffice/Office engine, and PostScript/EPS rasterisation needs Ghostscript (which is AGPL and belongs on a server, not in a shipped client bundle). These tools are clearly labelled as server-based before you use them.

Data deletion for server-based tools

When a conversion does run on our server, the file is sent over an encrypted connection (TLS), converted, made available for download, and then deleted. We do not retain your file contents after the job completes, and we do not use them to train anything.

Prove it yourself

You do not have to take our word for any of this. Turn off your Wi-Fi after a tool page has loaded and convert a file — the in-browser tools keep working, which is only possible if the conversion is happening on your machine. Every page also carries a live panel that instruments the browser's network calls (fetch, XHR, and beacon) and tallies the bytes actually sent. During an in-browser conversion it reads zero.

What we collect — and what we don't

  • WE DON'TThe contents of files converted in your browser — we never receive them.
  • WE DON'TWe do not sell personal data, and we do not use your files to train models.
  • WE DOIf you create an account: your email address and authentication credentials (passwords are hashed with scrypt, never stored in plain text).
  • WE DOIf you use the API or a server tool: job metadata (which conversion, when, success or failure) needed to run and bill the service.
  • WE DOPrivacy-respecting, aggregate analytics — only when configured, and never tied to file contents.

Full detail in the privacy policy.

Honest limitations

Client-side conversion is the right default, but it is not magic. These are the trade-offs we think you should know before you rely on a tool:

  • Everything runs in your device's memory, so a very large video can exhaust a browser tab where a data-centre server would not. Large files are bounded by your hardware, not ours.
  • Heavy engines (the FFmpeg core is around 30 MB, and the AI models are separate downloads) are fetched once on first use, then cached. The first conversion on a new device is slower than the ones after it.
  • PDF to Word recovers text and basic structure faithfully, but a complex multi-column layout may reflow — it is not a pixel-perfect visual clone.
  • Compression trades resolution for size. Image-heavy PDFs shrink dramatically; text-only PDFs were already small and will barely change.
  • OCR is highly accurate on clear printed text and unreliable on handwriting, glare, or blurred photos.

Supported browsers

The in-browser tools run in any modern browser. Because the work happens on your device, support and speed depend on your browser and hardware:

  • Chrome and Edge, desktop and Android — full support, including WebGPU acceleration for the AI tools where the device provides it.
  • Safari on macOS and iOS — supported; a few of the heaviest AI models fall back to running on the CPU.
  • Firefox — supported for all in-browser conversions.

How we make money

This matters, because "free" tools that process your documents have to earn a living somehow — and if the answer is your data, you deserve to know. It isn't. The in-browser tools are free and unlimited precisely because they cost us almost nothing to run: the conversion happens on your computer, not on servers we have to pay for.

We make money from a paid Pro plan (larger files, the developer API, and the server-powered conversions that genuinely need a machine in a data centre). That's it — we do not sell personal data, we do not sell your files, and we do not train models on them. See pricing for what's included.

Who operates File Transmute

File Transmute is built and operated by Unique Technology Group, based in Springfield, VA, United States. For security concerns — including responsible disclosure of a vulnerability — email security@file-converts.com. For privacy questions, email privacy@file-converts.com. You can read more about the project on the about page.