Basics
It inspects your IP address (VPN/proxy/Tor/datacenter status, ASN, geolocation, DNSBL listings) and your browser (canvas/WebGL/audio fingerprint, fonts, hardware, timezone, WebRTC and DNS leaks, automation tells), then cross-checks the two against each other for consistency and returns a trust score out of 100. For example, if your IP says you're in Germany but your system timezone and GPS both say California, that mismatch alone drags the score down, the same way a fraud-detection system at a bank or e-commerce site would flag it. Every signal that contributed is listed individually under the score, so you can see exactly which check moved the needle instead of just getting a number with no explanation.
A 0-100 score built the same way a fraud or anti-bot system would score a visitor, weighing IP reputation, browser fingerprint consistency, and leak signals together rather than any single check in isolation. 85 and above shows as "Trustworthy": a clean, consistent connection with no VPN/proxy/Tor flags and no fingerprint spoofing detected. 70-84 shows as "Suspicious": usually one or two signals disagree, like a VPN that passed most checks but still tripped a blacklist, or a browser fingerprint that looks slightly inconsistent. Below 70 shows as "Masked": multiple signals point to active anonymization or spoofing, most often a VPN/proxy/Tor exit combined with a randomized or blocked fingerprint. The exact band and which signals caused it are both shown on the result page, not just the number.
Yes, entirely free, no account, no API key, no credit card.
Privacy & Data
Your IP is hashed after 24 hours, and the full scan record is deleted after 30 days. See the full Privacy Policy for details.
No. We don't sell data, run ad-tech trackers, or share results with third parties.
No persistent tracking cookies. The only thing stored client-side is your dark-mode preference, in your browser only.
VPN, Proxy & Tor Detection
Some rebranded or newly provisioned VPN pool IPs aren't yet in our provider list or ASN deny-list. We continuously expand coverage, but no free IP intelligence source catches everything.
A small number of ISPs route traffic through infrastructure that resembles hosting or CGNAT ranges, which can trigger a false positive. If this happens consistently, it's worth flagging to your ISP.
IPs registered to cloud providers, VPS hosts, or known datacenter ASNs, as opposed to residential or mobile ISP ranges. This is one of the strongest signals fraud systems use.
Proxy-Check Tool
IP:PORT, USER:PASS@IP:PORT, IP:PORT@USER:PASS, IP:PORT:USER:PASS, USER:PASS:IP:PORT, and full URLs like socks5://user:pass@IP:PORT. Auto-detect handles a mixed list.Up to 100 per check, tested concurrently so a full batch takes seconds, not minutes.
Yes, HTTP, HTTPS, SOCKS4, and SOCKS5 are all supported, auto-detected per proxy or pinned via a
scheme:// URL.No. Proxy credentials are used in-memory for the duration of the live test only and are never logged or saved.
Technical
The GPS-vs-IP check compares your device's actual GPS/Wi-Fi location against where your IP says you are. It only runs if you explicitly opt in, we never request it automatically.
WebRTC (the browser API behind video calls and peer-to-peer connections) negotiates a direct connection by exchanging "ICE candidates," short descriptions of every IP address and port your device can be reached on. The problem is that WebRTC does this negotiation at the browser level, completely outside your VPN's network tunnel, so it can hand a website your real local network IP and, in many cases, your real public IP, even while the address bar still shows you're connected through the VPN. This is one of the most common ways a VPN gets unmasked without the user noticing, since nothing in the browser UI indicates it happened. ipscore.us tests for this directly by opening a WebRTC connection and reading back every ICE candidate it reveals, exactly the same technique a tracking script would use.
Not necessarily. Privacy-focused browsers and extensions (Brave's fingerprint randomization, Firefox's resistFingerprinting, CanvasBlocker, and most anti-detect browsers) deliberately inject a small amount of random noise into canvas and WebGL output specifically so the same scene renders a slightly different pixel hash on every run, which defeats canvas fingerprinting as a tracking technique. ipscore.us renders the same canvas/WebGL scene twice in a row and compares the two hashes: real, unmodified hardware produces an identical hash both times, while noise-injected rendering produces two different hashes. We surface that as an informational signal (hardware fingerprint appears to be randomized) rather than an automatic penalty, since it's evidence of a privacy tool working as intended, not evidence of something suspicious.