How to Edit Video Without Uploading It Anywhere

Most online video editors start the same way: you pick a file and a progress bar crawls across the screen while it uploads. If the footage is a client's, a patient's, a child's or simply yours, you may want to edit video without uploading it at all. This post explains why most browser editors upload first, why a browser can now do the whole job on your own machine, and a test you can run on any editor, ours included, to see whether your video leaves.

Why most online editors upload first

In a cloud editor, the heavy work happens on the company's servers. Your browser shows the timeline and sends your instructions, and the servers decode the footage, render the preview and produce the export. That design needs your file on their machines before anything can happen, which is what the upload bar is.

There is nothing sinister about it; it was the only practical way to edit video in a browser for a long time. But it does mean a copy of your footage exists somewhere else, under that company's terms and for as long as their retention says. Our post on a free CapCut alternative that keeps footage local walks through what one such set of terms actually says.

Why a browser can now edit locally

The change is a browser feature called WebCodecs. MDN describes it as letting web apps encode and decode video and audio in the browser, efficiently, with hardware acceleration and control over individual frames, and it names browser-based video editing among the uses it is built for. MDN also notes that individual devices and browsers may only support some codecs.

In practice, that means a web page can read a video file straight off your disk, decode its frames on your own processor and graphics hardware, and encode a finished MP4 without a server in the loop. The file never has to go anywhere. Whether it does is a choice the editor makes, and you can check which choice it made.

How to check whether an editor uploads your video

You do not have to trust anyone's marketing for this, including ours. Use a large file, a few hundred megabytes or more, so an upload has nowhere to hide.

1. Look for an upload bar. It is the most honest signal there is. If importing shows a percentage climbing towards 100, the file is going somewhere.
2. Open the browser's network log before you import. In Chrome, open DevTools and click the Network tab. Chrome's documentation notes that DevTools only logs network activity while it is open, so open it first, then import.
3. Watch what appears during the import. An upload shows up as requests that start the moment you import and keep running until the progress bar finishes, often one long request or a stream of repeated ones to the same server. A handful of small requests for sign-in or analytics is normal and tells you nothing either way. Chrome's Size column shows what was transferred over the network for each request, which mostly reflects what came back, so do not rely on it alone to measure an upload.
4. Check what your computer actually sent. On a Mac, open Activity Monitor, choose the Network tab and watch the Data sent total at the bottom while you import. On Windows, Resource Monitor's Network tab shows how fast each process is sending. Importing a two-gigabyte file into a cloud editor moves that figure by roughly two gigabytes; a local editor barely moves it.
5. Watch the export. A local editor renders on your machine and saves the file directly. A cloud editor renders remotely, and you wait for a download.
6. Find the opt-ins. Many editors that work locally still send something for particular features, usually AI ones. Read the editor's own description of which features send what, and when.

Running the test on flshy

flshy is built to pass that test, so run it. Here is what you should and should not see.

You should see small requests: signing in, and Google Analytics, which runs in the editor as well as on this site. Our privacy policy says what it records, and that it never receives your footage, your projects or anything on your timeline. The first time you generate captions on your device, you will also see the speech model download, about 42 MB. After that it is cached, and on-device captioning works with no network at all, as the captions guide explains.

You should not see anything the size of your video. Import, editing, preview and export all happen on your machine, and the video files themselves are never uploaded. There is no upload bar, because there is no upload.

Three features do send something, and only when you ask: cloud captions and AI Highlights each send the clip's audio to a speech-to-text service, and the Director chat sends a summary of your project, meaning its name, your track and clip layout and your media file names, to a hosted Claude model. If you want nothing to leave the machine, leave those three alone.

If the test ever shows something we have not described here, tell us at support@tryflshy.com. That would be a bug, and we would want to know.

The trade-offs of editing locally

Keeping the footage on your machine is not free of costs, and it is worth knowing them before you start.

• Nobody else has a copy. Projects are stored in your browser. Clear that storage and the project is gone, and flshy cannot recover it, because we never had it.
• Your machine does the work. Export is hardware-accelerated H.264 up to 4K, rendered on your own hardware, so its speed is your computer's speed.
• It needs a computer. flshy runs in a desktop or laptop browser, best in Chrome or Edge, and not on phones.

Editing without uploading, start to finish

Every step below runs on your own machine.

1. Sign in and bring the file in. Make a free account, then drag your video onto the timeline. Your browser reads it off your disk.
2. Cut the dead air. Silence removal analyses and cuts the audio locally; nothing about it touches the network. See remove silence from a recording.
3. Caption it on the device. On-device transcription is the default, in 14 languages plus auto detect, and every cue lands as an editable text clip.
4. Reframe it. Auto Reframe decodes and analyses frames in your browser to track the subject for a 9:16, 1:1 or 4:5 crop. See turn a landscape video vertical.
5. Export. H.264 MP4 up to 4K, with no watermark, saved straight to your computer.