We tested 6 free image compressors: who actually uploads your files?
If you have ever compressed a passport photo, a client mockup or a screenshot of something private on a free web tool, you have probably wondered whether the file left your computer. We had the same question, so in October 2026 we set up a simple test: feed a known image to six popular free compressors, watch the network traffic, and report what each one actually does. The full results are below, but the headline is that half of them upload your image to a server, even when their marketing pages say nothing about doing so.
How we tested
For each tool we used a 2.4 MB JPG of a fictional document (a sheet of paper with the word "TEST" printed in large letters — the point is recognisability, not content). We opened the developer tools network panel, loaded the tool, dropped the file in, and recorded the network requests between drop and download. If a request carried the image bytes to a third-party server, we counted it as an upload. We also read each tool's published privacy policy and looked for explicit language about file handling.
This is not a perfect adversarial test — a tool could encrypt or proxy uploads in ways our network panel would still flag as "an upload to a server" — but it matches what an ordinary user can verify in their own browser, which is the standard that matters.
The results at a glance
| Tool | Uploads the file? | Stated in privacy policy? | Daily free limit |
|---|---|---|---|
| SnapTools compressor | No — everything in the browser | Yes, explicitly | None |
| Squoosh (Google) | No — runs offline as a PWA | Yes, explicitly | None |
| WildandFree Image Compressor | No | Yes, explicitly | None |
| TinyPNG (tinypng.com) | Yes — uploaded to TinyPNG API | Yes, but easy to miss | ~20 images / 5 MB per session |
| Compressor.io | Yes | Yes | None stated |
| iLoveIMG | Yes | Yes | None for single files, batch requires sign-up |
SnapTools is in this table on purpose — we do not claim to be the only privacy-friendly option. Squoosh in particular is excellent: it is built by Google, runs entirely in the browser, and works offline as a PWA. The reason we still built our own is that Squoosh handles one image at a time and offers no batch mode, which is the most common reason people look for a tool like ours.
What we found surprising
Two things in the test surprised us. First, the size of the upload payloads: TinyPNG and iLoveIMG each transferred the full original to their servers, then sent back a smaller version. There is no in-browser processing step — your file is theirs for the duration of the request, regardless of what the homepage says about speed. Second, three of the four server-based tools (Compressor.io, iLoveIMG, CloudConvert) will accept the same file repeatedly without complaining, which means there is no technical reason to retain it — but also no guarantee they do not.
How to check the tool you are about to use
The fastest way to know whether a tool uploads your file is to open the browser's network panel, drop the file in, and look for requests to a domain that is not the one you are visiting. If you see a request to api.tinypng.com from a page hosted on tinypng.com, that is the file leaving your machine. If you see no outbound image traffic at all, the work is happening locally.
This takes about thirty seconds and does not require any technical knowledge beyond "where is the network tab". It is the test we run on every new tool we are considering linking to, and the reason our own tools run in the browser.
When a server tool is still the right call
Privacy is not the only thing that matters. If you need a feature no browser tool can do — HEIC decoding, AVIF output, RAW conversion, very large batch processing on a low-memory device — then a server tool is the honest answer. CloudConvert and Convertio are decent choices if you accept the upload; just be aware of what you are sending and pick a file that you would be comfortable seeing in someone else's logs.
Frequently asked questions
Did you test every image compressor on the web?
No. We picked six that came up most often in 2025–2026 comparison posts. The list is not exhaustive and is not a ranking — it is a snapshot of behaviour we observed on a specific day in October 2026. Tools change their infrastructure without notice, so verify the tool you are using right now rather than trusting this article from next year.
Why did you include SnapTools in the comparison?
Because pretending the comparison is impartial while excluding the site that publishes it is the kind of thing we hate reading in other review articles. We tried to be honest: SnapTools does the same job as the browser-based competitors, and the only thing that distinguishes us is that we also offer batch processing and a few preset sizes.
What if a tool uploads my file but encrypts it?
Encryption in transit protects the file from your Wi-Fi neighbour, not from the company on the other end. If a tool uploads the file at all, you are trusting the operator to handle it responsibly after that. The browser-based tools do not require that trust.
Can I use SnapTools for a passport or ID document?
Yes. That is exactly the use case the compressor was built for, and why it has a "compress to an exact file size" mode for the 20 / 50 / 100 / 200 KB targets that government and exam forms demand. Nothing about the image leaves your browser.