Claude Risk Signal Check: What Region Your Browser Looks Like
The page collects 10 browser signals automatically and computes a weighted reference score. Every signal expands to its raw value — transparent enough to recompute by hand.
Straight talk first: this is a signal self-check, not an official verdict. Claude does not publish its regional decision logic; the weights on this page are our own estimate from public browser knowledge. A high score does not mean a guaranteed block, and a low one promises nothing.
Reading browser signals
Result-area disclosure: this score only reflects how strongly your browser exposes China-region-related signals. It has no connection to any official Claude system, and this page neither provides nor encourages means of evading regional restrictions.
The 10 signals in detail
Tap any signal to expand its raw value and reasoning. “Unavailable” = the browser does not expose that API — marked honestly, excluded from scoring.
Scoring formula (transparent, recomputable)
| Signal | Weight | Raw | Weighted |
|---|
Total = Σ(raw × weight), normalized over available weights when signals are unavailable. Raw scores come in three steps: 0 (no China-region signal) / 50 (neutral) / 100 (strong signal).
Why this tool exists
A friend asked me last week: "My network setup hasn't changed, but Claude sometimes loads instantly and sometimes spins forever — is it singling me out?" I couldn't answer, because Claude does not publish its decision logic — and anything claiming to be an "official detector" deserves suspicion. What I could do was lay the browser's own exposed signals on the table: what your timezone, language, fonts and renderer look like to others, at a glance.
So the tool's position is a health check, not a prescription. Signal self-check means: it tells you what your browser looks like, not how to disguise it — changing your system language or timezone is your own freedom, but using fingerprint spoofing to dodge service terms is a different matter this page does not touch.
How the 10 signals were chosen
Selection rule: everything readable passively by page JS with zero permission prompts (keyboard layout excepted — it needs a new API most browsers lack). Timezone and language carry the top weights (20% each) as the most common regional hints; UTC offset and date format cross-validate the timezone; the font fingerprint (Canvas-measured CJK glyph widths) reveals whether Chinese font libraries are installed; the WebGL renderer string often carries regional traits on domestic OS builds and common GPUs; touch, screen and DNT weigh little and serve as auxiliary context.
| Input | No input required: signals collect automatically on load; the re-run button redoes the pass (results should be identical — the signals themselves are deterministic) |
|---|---|
| Output | 0–100 reference score + three-tier interpretation (low/mid/high) + per-signal raw value, tier and weighted contribution + the formula table |
| Weights | Fixed: timezone 20%, language 20%, UTC offset 10%, date format 10%, fonts 10%, WebGL 10%, screen 5%, touch 5%, keyboard 5%, DNT 5%; no user-adjustable parameters |
| Processing | Browser-local: Intl.DateTimeFormat / navigator.* / Canvas font metrics / WebGL parameter reads, all computed locally, zero uploads zero retention (the site-wide analytics at page bottom never touches fingerprint data) |
| Bounds | Fixed set of 10 signals; no files, no uploads, no storage quotas involved |
| Known limits | Not an official verdict (Claude publishes no decision logic); unavailable signals are excluded from scoring rather than counted as low risk; results are constant per browser and device; private-browsing modes may alter some signals |
FAQ
Can this score predict whether Claude will block me?
No — treat it as context only. Claude does not publish the specifics or weights of its regional decisions. The 10 signals and weights on this page are our own estimate built from public knowledge of browser characteristics, not a reverse-engineering of any official system. A high score does not guarantee a block, and a low one promises nothing. Read it as a health check of how China-region-like your browser looks.
Is any of this data uploaded?
No. Every signal is collected and computed inside this device's browser; the page makes zero network requests of its own (aside from the site-wide analytics at the bottom, which never touches fingerprint data) and results are only shown on the page. Nothing is logged, sent back or stored.
Why do some signals show unavailable?
Browsers gate some interfaces behind privacy rules: the keyboard layout API (navigator.keyboard) exists only in recent Chrome and needs permission, while Firefox and Safari do not ship it at all. Signals that cannot be read are honestly marked unavailable — they are excluded from scoring and never guessed. Better one signal fewer than one fabricated.
What is the scoring formula?
Each signal gets a raw score of 0, 50 or 100, multiplied by its weight — timezone 20%, language 20%, UTC offset 10%, date format 10%, font fingerprint 10%, WebGL renderer 10%, screen 5%, touch 5%, keyboard layout 5%, DNT 5% — then normalized over the available weights to 0-100. The formula and per-signal scores are shown in the results so you can recompute by hand.
My score is high — what should I do?
Look at which signals pulled it up: timezone and language carry the most weight. System language and timezone follow your OS settings, and the browser language list is adjustable in browser settings. Note that this tool only performs a signal self-check — it does not provide or encourage any means of evading regional restrictions; Claude’s availability policies are defined by its official terms.
Isn’t this a fingerprint-tracking tool?
It’s the opposite direction. Tracking tools harvest your fingerprint, upload it and correlate you across sites; this page lays out which signals your browser exposes for you to see, entirely locally with zero uploads. Knowing what your browser looks like is the first step in understanding fingerprinting itself.