What specific fingerprinting tests (e.g., canvas, WebGL, timezone) does PrivacyTests.org include and which browsers fail them most often?
Executive summary
PrivacyTests.org is an open-source test suite that runs automated browser checks for fingerprinting resistance, site partitioning, and related privacy behaviors; its code and test pages are published on GitHub and the project publishes per-browser results and commentary [1][2][3]. The public reporting highlights specific vectors—third‑party cookies, screen and font fingerprinting, and other canvas/feature leaks—while noting that different browsers show different strengths and weaknesses [4].
1. What PrivacyTests.org actually tests: the high level inventory
PrivacyTests.org runs a suite of automated browser privacy checks described as “fingerprinting resistance, partitioning between websites, etc.” and executes a set of test pages (fingerprinting.html, https.html, misc.html, supercookies.html) that exercise many browser surfaces and storage mechanisms [2][5]. The project’s site presents those rendered results for human consumption and the codebase is openly available on GitHub [3][1].
2. The concrete fingerprinting vectors exercised (canvas, fonts, cookies, and more)
The tests exercise classical fingerprinting surfaces such as canvas rendering differences and other client-side script-derived identifiers (the project integrates the browser-privacy tests hosted at arthuredelstein and related pages) and it explicitly evaluates “screen fingerprinting” and system font exposures; PrivacyTests.org’s commentary notes Brave’s recent protection for screen fingerprinting and a Brave Nightly mitigation for system font fingerprinting [5][4]. The suite also includes checks that touch cookie/partitioning behaviors (including third‑party cookie blocking), cache and storage behaviors labelled in the logs as favicon/cache/font/CSS entries, Web SQL and TLS/session indicators, and other “supercookie” or storage‑based tracking channels [5][2].
3. Which specific browsers fail which tests, according to available reporting
Public notes from PrivacyTests.org call out differences among mainstream browsers: Safari and Tor Browser are reported as protecting against “this type” of fingerprinting in a highlighted issue (specifically in the context PrivacyTests.org discussed), while Brave Nightly is explicitly named as passing screen fingerprinting and adopting a randomized user-installed font exposure to mitigate system font fingerprinting [4]. The project is framed as comparing “major web browsers” so results reference Chrome, Firefox, Brave, Edge, Opera and variants, but the provided reporting does not include a complete, current ranked table in these snippets — only selective examples and change notes [3][4].
4. Limitations, methodology notes and replication caveats
Because the test harness and rendering are open-source, the exact suite run against any browser depends on the test scripts and platform flags (mac/linux/ios/android) and the project’s scripts directory documents those parameters; replication attempts have encountered practical issues such as unsupported features, TLS errors and local rendering bugs noted in issue logs, which complicates straightforward cross‑browser aggregation [2][5]. PrivacyTests.org itself acknowledged a results error in an early issue (a misreport that Safari did or did not stop third‑party cookies), underscoring that snapshot results can change and must be interpreted with caveats [4].
5. Practical takeaway: what can be said with confidence
PrivacyTests.org actively exercises a broad set of fingerprinting vectors—canvas/screen rendering, font enumeration, cookie and partitioning behaviors, storage/supercookie channels, Web SQL and TLS-level indicators—using automated test pages hosted in its codebase [2][5][1]. Public snippets of the project’s reporting highlight that browsers differ: Tor and Safari are called out for protecting against certain leaks, while Brave has shipped mitigations for screen and font fingerprinting; however, the sources provided do not deliver a full, up‑to‑date ranked list of “which browsers fail most often,” and replication logs show variability that prevents declaring a definitive failure ordering from these excerpts alone [4][5][3].