Skip to content

Browser Fingerprint Test

Everything is read and hashed in your browser. Nothing is sent, stored or counted, so this page cannot tell you how rare you are. The bit figures come from published studies.

Reading your browser signals...

Reading your browser signals...

Learn more

Guide and answers to common questions about Fingerprint Test

What browser fingerprinting is

Browser fingerprinting is a way to recognise a browser without storing anything in it. A script on a web page reads many small facts that every browser gives away: the screen size, the time zone, the languages, the graphics card, the fonts that are installed, and how the machine draws a picture or processes a sound. Each fact is shared by many people, but the combination of a dozen of them can be rare enough to work as a label for one browser.

This page reads those facts in your browser, shows each value, and says how identifying that kind of signal is according to published research. It is the same reading that an advertising or anti-fraud script would do. Nothing you see here leaves your device.

How it differs from cookies

A cookie is a small file that a site puts in your browser and reads back later. You can see it, delete it and block it. A fingerprint is not stored in your browser at all. The site works it out again on every visit from what your browser reveals, so clearing cookies does not remove it, and a private window usually has the same fingerprint as a normal one.

A fingerprint is also less exact than a cookie. A cookie with a random number is unique by design. A fingerprint is a guess that two visits come from the same browser, and it breaks when the browser updates, the window is resized or a new font is installed. Trackers therefore often combine it with other signals such as the IP address, and use it to bring back an identity after the cookies are gone.

What the studies found

The first large study was the EFF Panopticlick project. Peter Eckersley published it in 2010 as "How Unique Is Your Web Browser?", based on about 470,000 browsers that visited the test site. Among the characteristics he measured, the list of plugins carried about 15.4 bits of identifying information, the list of fonts about 13.9 bits, the user agent about 10.0 bits, the screen resolution and colour depth about 4.83 bits and the time zone about 3.04 bits. Whether cookies were switched on carried only about 0.353 bits, because nearly everyone has them on. A bit here is a halving of the crowd: ten bits separate one browser from roughly a thousand.

In 2016 Laperdrix and colleagues published "Beauty and the Beast" about the AmIUnique site. They added newer signals such as canvas rendering and WebGL, and confirmed that fingerprints could be used to tell browsers apart, also on mobile devices. In 2018 Gómez-Boix, Laperdrix and Baudry published "Hiding in the Crowd", measuring more than two million browsers on a popular French website. They found that a much smaller share of browsers were unique than the earlier studies suggested: around a third of desktop fingerprints, and fewer on mobile. The reasons include that plugins and fonts are no longer exposed the way they were in the Flash and Java era, and that browsers had become more alike.

The lesson is not that fingerprinting does not work. It is that its power depends on the site, the visitors and the year. The figures on this page are from those papers and describe the browsers of their time. Where a number was not clearly published, the page gives only a level: high, medium or low.

Why bits cannot be added up

It is tempting to add all the bits into one total, but that would be wrong. Signals overlap. A user agent that says "iPhone" already tells most of what the screen size and the touch points say. A fixed set of fonts follows from the operating system. Adding correlated signals counts the same information several times.

For that reason the page shows the strongest signals rather than a score, and it does not say whether you are unique. Exposy does not count visitors, so it has nothing to compare you with. Only a study of many browsers can say how rare a combination is.

The signals this page reads

Every item below can be read by any web page without a permission prompt. The page hashes large values such as canvas and audio so that you can compare two runs without a long string.

  • Canvas 2D: the page draws text and shapes on a hidden canvas and hashes the pixels. Small differences in fonts, drivers and anti-aliasing change the result.
  • WebGL: the name of the graphics card and driver, and a hash of the limits and extensions of the graphics stack.
  • Audio: a sound is rendered offline and the samples are hashed. Differences in the audio code give slightly different numbers.
  • Fonts: about fifty common fonts are tested by measuring the width of a text. Installed programs add fonts, which makes a list rare.
  • Screen: size, available size, colour depth, window size and pixel ratio.
  • Software: the user agent, the client hints of Chromium browsers, plugins and the PDF viewer.
  • Settings: languages, time zone, cookies, Do Not Track, Global Privacy Control and display preferences such as dark mode.
  • Hardware: the number of processor threads, the device memory, the touch points and the platform string.
  • Storage: which storage and worker interfaces exist.
  • Devices: only the number of cameras, microphones and speakers. The page never asks for camera or microphone access, so it sees no names.

Noise and randomisation

Some browsers defend against fingerprinting by adding small random changes to canvas and audio results. To test this, the page draws the same picture twice and compares the two hashes. If they differ, your browser adds noise, and the canvas signal no longer works as a stable label. The audio test does the same.

This check has a limit. Some browsers add noise that is the same for a whole session on one site, so two draws in a row look identical. Those browsers are not caught by this page, and the result then says only that the hashes matched.

How to reduce your fingerprint

There are two approaches, and they work against each other. One is to look the same as everyone else. The other is to look different every time. The first works only when a large group shares a setting.

  • Tor Browser aims to make all its users look alike, with fixed window sizes, fonts and time zone. It also asks before a page may read canvas data.
  • Firefox has a setting, privacy.resistFingerprinting, that reports common values for several signals. It can break some sites and change how pages look.
  • Brave adds small random changes ("farbling") to canvas, audio and some other readings, so a fingerprint changes from site to site.
  • Safari limits some of the signals in its tracking protection. How much depends on the version and the settings.
  • Extensions that block scripts or trackers stop many fingerprinting scripts from running, but an extension is itself visible and can add a signal. Use few.
  • Keep the browser updated and avoid unusual settings, fonts and extensions. The more your setup differs from the default, the more it identifies you.

Why Exposy does not store fingerprints

A fingerprint hash identifies a browser, which makes it personal data in many privacy laws even without a name. A test that stored hashes in order to count how unique you are would build the kind of profile it warns about. Exposy therefore stores nothing, sends nothing and counts nothing for this page: the code runs in your browser, and the figures are taken from the papers.

To check it, open the network tab of your browser tools while the page runs. You will see no request carrying any of the values.

Frequently asked questions

Does a private window stop fingerprinting?

No. A private window removes cookies and history when it closes, but it still gives the same screen, fonts, graphics card and time zone. Some browsers add extra protection in private mode, but do not count on it.

Does a VPN change my fingerprint?

No. A VPN changes your IP address, not the signals read inside the browser. A site can still recognise the browser, and a time zone or language that does not fit the VPN country can even stand out.

Is it illegal to fingerprint a browser?

It depends on the purpose and the place. In the European Union, the ePrivacy rules and the GDPR generally require consent for tracking by fingerprint, with exceptions for things such as security and fraud prevention. This page is not legal advice.

Why do two runs give a different canvas hash?

Your browser most likely adds noise on purpose, as some privacy browsers and extensions do. An unstable graphics driver is a rare other cause. If the hash is the same each time, the browser either adds no noise or adds noise that is fixed for the session.

Are these bit figures about me?

No. They describe the typical identifying power of a signal in the studies, using the browsers of that time. They do not say how rare your own browser is.

Is anything sent to Exposy?

No. The signals are read in your browser and hashed there. Nothing is uploaded, saved or counted. The Copy as text button only puts the results on your clipboard.