How to check whether an online converter uploads your file
Every online converter says your files are safe, including the ones that upload everything. You do not have to take anybody's word for it, ours included. These three checks use only what is already in your browser, work on any converter, and take a few minutes the first time.
Updated
Before you start: what an upload looks like
A converter can do the work in one of two places. A server-side converter sends your file to a computer it controls, converts it there and sends the result back. A browser-based converter downloads its code once and converts the file inside your tab. Both can look identical on screen. The difference shows up in one place your browser lets you watch: the requests the page sends.
The pattern to look for is a request that carries your file out. In practice that is almost always a POST (sometimes PUT) made by the page's scripts, which developer tools label Fetch/XHR, whose request body is about as large as the file you chose. It may go to the site's own address, to a separate API or upload host, or to a storage bucket. Large files are often split into several smaller requests, so add those up. A file can also leave through a WebSocket (the WS filter) or a plain form submission (the Doc filter), so glance at those too.
What is not an upload: GET requests that download scripts, styles, fonts or WebAssembly files. Those can be large, but the size is in the response, travelling to you. Small POST requests of a few hundred bytes are usually analytics or error reporting; they send data about your visit, which you may care about, but they cannot contain a photo of several megabytes.
Check 1: watch the Network tab
This is the main check. It shows every request the page makes, with its method, its destination and its size, and it is built into every desktop browser.
Open the Network tab
- Chrome and Edge: press F12 or Ctrl + Shift + I (Cmd + Option + I on a Mac), then choose the Network tab. Edge uses the same DevTools as Chrome, so the steps below are identical.
- Firefox: press Ctrl + Shift + E (Cmd + Option + E on a Mac) to open the Network Monitor directly, or F12 and then the Network tab.
- Safari:first turn on the developer tools once: Safari, Settings, Advanced, then tick "Show features for web developers" (on older versions, "Show Develop menu in menu bar"). Then press Cmd + Option + I and choose the Network tab.
Run the test
- Load the converter page and let it finish loading. Keep the Network tab open while you do.
- Turn on Preserve log (Chrome, Edge, Safari) or Persist Logs (Firefox, in the Network settings menu), so a redirect or a page change after the conversion does not wipe the evidence.
- Clear the list with the clear button, so only what happens from now on is shown.
- Choose or drop a file you recognise by size, for example a 3 MB photo. Convert it and download the result, exactly as you normally would.
- Filter on Fetch/XHR (Firefox calls it XHR) and sort or scan by method. Click any
POSTorPUTand open its Payload (Chrome, Edge) or Request (Firefox) panel, or the request section of the Headers view in Safari. If the body is your file, or theContent-Lengthrequest header is close to your file's size, the file was uploaded. - Check the other filters once (All, WS, Doc) for anything sent while you converted.
On a server-side converter you will see that upload the moment the file goes in, often followed by repeated status checks and a download of the result. On a browser-based converter you will see, at most, the page fetching more of its own code (a decoder or an encoder it loads only when needed) and then nothing while the conversion runs.
On this site, the requests you can expect after adding a file are downloads from this same domain: the code for the format you picked, the AVIF encoder (a WebAssembly file) if you export AVIF, or the video engine on the video tools. None of them carries anything of yours.
Check 2: go offline after the page has loaded
This one is a useful second opinion, as long as you run it fairly. The idea: once the page and everything it needs are loaded, a converter that works in your browser does not need the network to convert a file. A converter that sends the file to a server cannot finish without it.
- Load the converter page with your connection on, and convert one file of the type you want to test. That makes the page fetch any decoder or encoder it loads on first use.
- Go offline without closing the tab: switch on airplane mode or turn off wifi and unplug any cable. In Chrome and Edge you can also pick Offline from the throttling menu in the Network tab, which cuts only that tab off.
- Do not reload the page. Convert a second file of the same type and download the result.
If the second conversion completes, the work happened on your device. If it stalls, shows a connection error or waits forever, the converter needed a server for it.
Read the result honestly in both directions. A page without a service worker (this one has none) cannot be reloaded offline, so a reload failing proves nothing. And a browser-based converter that has not yet fetched a decoder, for example because you skipped step 1 or switched to a format it loads separately, will also fail offline without having uploaded anything. That is why the Network tab in check 1 is the authority and this test is the confirmation.
Check 3: read the CSP connect-src header
A site can send a Content-Security-Policy header that tells your browser where the page's scripts may connect. The connect-src directive covers fetch, XHR, WebSockets and beacons. Your browser enforces it, not the site: a request to a destination the policy does not list is blocked before it leaves.
- In the Network tab, reload the page and click the first request, the document itself (filter on Doc if the list is long).
- Open its Headers panel and find
content-security-policyunder the response headers. From a terminal,curl -sI https://example.com/with the converter's address shows the same headers. - Find
connect-srcin that header. If it is missing, look fordefault-src, which applies in its place.
How to read what you find:
connect-src 'self', optionally withblob:anddata:, means scripts can only connect to the site's own address. No third-party upload host, no outside storage.- A list of other domains (an API host, a storage service) means the page is allowed to send data there. That is not proof it does, but check 1 will show whether it does.
- No CSP, or only a
Content-Security-Policy-Report-Onlyheader, means nothing is restricted: a report-only policy reports violations but does not block them.
The limit of this check, stated plainly: 'self'still allows a request to the site's own server, so a strict policy rules out uploads to anyone else, not uploads to the site itself. Check 1 is what rules out the second.
This site sends connect-src 'self' blob: data:. You can confirm it with curl -sI https://imageconvert.tools/.
How this site shows it while you convert
To save you opening DevTools every time, the converter on this site shows a small counter once you add files. From the moment you add files, it shows how many network requests the page has made, how many bytes were downloaded, and whether every one of them went to this site. When a decoder or encoder is fetched, it names it. If any request went to another site, it says so, and changes its look.
The numbers come from your browser's Resource Timing API, the same measurements developer tools are built on, not from anything the site reports about itself. It can only count what the page's own performance timeline records, so treat it as a convenient summary and the Network tab as the full picture. In browsers without the API the counter does not appear at all rather than showing a guess.
What this does not claim
Being straight about the edges is the point, so here is what is outside the promise.
- Your files stay on your device during conversion. What you do with the result afterwards is between you and wherever you send it.
- The site is served over the internet like any other, so your browser's request for the page itself is visible to your network and to the host, exactly as it is for every website you visit. What is not visible is the content of your files, because they are never part of a request.
- The site does not work offline from a cold start: there is no service worker, so the page and any decoder or encoder must be loaded first, as described in check 2.
- The WebAssembly engines are hosted here rather than on a public CDN. That is deliberate: a CDN would mean a third party observing which visitors convert video. It also means the download is slower than it could be.
- No account, no cookie wall, and no analytics that needs consent. If a cookie banner ever appears on this site, something has gone wrong.
When a server-side converter is the right tool
Server-side converters are not dishonest for working the way they do, and many delete files promptly. Their architecture simply means the file has to reach their machine before anything can happen to it, which is why the checks above matter. It is also the honest reason to use one sometimes: if you need a format no browser can decode, or a file too large for your device's memory, a server is the right place for it. ImageConvert.tools covers the common formats, on your own hardware, without the upload.
The mechanics are written out in full on how this converter actually works, down to which decoder handles which format and what happens to your EXIF data.
Sources
- Chrome DevTools: Inspect network activity
- Chrome DevTools: Network features reference (filters, throttling, Offline)
- Microsoft Edge DevTools: Inspect network activity
- Firefox DevTools: Network Monitor
- Apple: Enabling Web Inspector and developer features in Safari
- MDN: Content-Security-Policy connect-src
- MDN: Content-Security-Policy-Report-Only
- MDN: Resource timing
- MDN: Service Worker API