Not much on its own. A timezone or language that doesn’t match the exit IP mostly tells a site you’re on a VPN, which you’re fine with. It only starts to narrow you down when the combination is rare, for example an uncommon language plus a timezone few people on that exit server use, or when WebRTC leaks your real IP.

What actually lets a site recognize you again is the stuff that stays the same when your IP changes: cookies and logins first, then canvas, WebGL, fonts, screen size and similar. So if you only care about recognition, that part of the report matters more than the VPN mismatch part.

Fair question. No, correct matters more. “At this speed” just meant I couldn’t have built it alone in my spare time this fast.

For a page like this, being wrong is the real failure: if it says you’re fine when your WebRTC or time zone is leaking, that’s worse than no page at all. That’s part of why the code is public under MIT: https://github.com/yt1988/who-are-you. If you or anyone spots a check that reports the wrong thing, an issue with your browser and VPN setup is the most useful thing you can send me, and I’ll fix it.

Correction to my post: I wrote that the page itself isn’t open source. That was wrong. The source is public under MIT at https://github.com/yt1988/who-are-you, so you can check what it collects instead of taking my word for it. Sorry for the mix-up.

Fair enough, and thanks for the straight answer rather than just a downvote. Point taken: I won’t post it to more general communities here. If I share anything again it’ll be somewhere that’s explicitly fine with AI-assisted projects.

Yep, webkay and pages like it are where the idea comes from, nothing new about the basic checks. What I tried to add is the comparison part: does your timezone and language match your exit IP, does traffic leave through a different exit when a proxy splits it, and which of your values are rare. If the older pages already cover what you need, no reason to switch.

Honestly, not at this speed. I can read and fix the code, and I decide what the page checks and how it handles data, but a lot of it was written with an AI assistant. That’s why I said so in the post. If that’s a dealbreaker for you, I get it.

Thanks! If you want quick wins: the biggest signals are usually WebRTC leaking a local or real IP behind a VPN, a timezone or language that doesn’t match your exit IP, and a very unique canvas/font fingerprint. Firefox with resistFingerprinting, or the Tor/Mullvad Browser, cuts most of that down.

Fair point, and that’s why the post says AI helped. I used it for a lot of the code and the English text. What the page checks, and how it handles data (checks run in the browser, no IP or full fingerprint kept, only anonymous per-value counters that ‘Don’t count me’ or GPC turns off), was my call, and so is any bug you find in it.

Fair question. The account is new (made it this week), but it’s a real person: I’m the author of the page and said so in the post. I’m not running other accounts here, and I’ll stick around to answer questions or take criticism. If posting your own tool in c/webdev feels off-topic, tell me and I’ll keep it to comments.