Am I Tracked? is an open-source browser privacy demonstration. It shows what a normal web page can learn before and after you grant a permission, with each result explained in plain language.
The passive stage reads browser-exposed signals such as your device and operating system, browser and version, languages, timezone, screen, hardware hints, battery support, WebGL, and canvas/audio rendering. These signals are combined into explanations and confidence labels; they are not a claim that every signal uniquely identifies you.
The test may also attempt a client-side WebRTC public-IP check and a direct lookup that turns that IP into an approximate city and ISP. Hardened browsers, extensions, and network settings can block these readings, which is shown as unavailable or blocked rather than guessed.
The score is a weighted summary of the scored signals this page could read. Signals have different weights, so the percentage is not simply the share of readable fields. It is not a safety rating, a probability of being tracked, or a comparison with other visitors. Optional permission demos can change the score.
Tracker checks fetch six fixed endpoints with credentials and referrers omitted, without executing the scripts. Reachable means fetch completed; an opaque response does not reveal its HTTP status. Not reached means a request failed while an existing asset on this site completed as a connectivity control. A timeout, or a failure without a completed control, is inconclusive. Even with a working control, failure can be a network error or endpoint outage; it is never proof of blocking or tracking. The control only checks this site's connectivity, not general internet access.
The experiment does not rerun the reveal, IP lookup, fingerprint or permission demos. The per-endpoint table labels comparisons changed, unchanged or inconclusive. A changed result cannot establish what caused it or what any tracker retained. Cookie protection can leave credential-free endpoint fetches unchanged; this test does not measure third-party cookie blocking. GPC is displayed separately: an absent API means support is unknown, and a reported opt-out does not prove compliance.
The baseline stays in tab-local sessionStorage for up to 30 minutes, including across reloads, until you choose Finish & clear. It stores no IP address, fingerprint, permission or behavior readings. Browser session restore can restore tab storage; expired or incompatible snapshots are discarded. Storage denial falls back to memory with a warning that reload can lose the baseline.
Precise location, camera and microphone devices, clipboard, and notifications are requested separately and only after you choose to continue. The test does not record camera or microphone content. It also demonstrates tab attention, pointer, click, scroll, and dwell signals that pages can observe without a permission prompt.
The reveal can make network requests to the WebRTC STUN service, the disclosed IP-geolocation provider, and tracker endpoints to show whether those resources are reachable from your browser. These requests are part of the demonstration and can be blocked by your browser or extensions.
Raw test signals are computed in your browser and are not sent to an Am I Tracked backend. A localStorage value is used only for the returning-visitor demonstration, so it stays on your device. Third-party requests and anonymous analytics are described in the privacy policy.
Sharing is optional, and you can preview the text before copying. The reveal summary contains readable-signal and tracker counts; the experiment summary contains before/after and comparison counts. Both include a public test link and omit your IP address, location, fingerprint and individual readings. Nothing is posted on your behalf.
Browser APIs differ by browser, operating system, extensions, and network. A blocked or unavailable value is a real result, not an inferred value. You can inspect the collection logic, explanations, and tests in the public repository:
View the Am I Tracked source on GitHub ↗
If a browser change makes a result or explanation inaccurate, you can report a correction in GitHub Issues ↗.