Hi, this is a particularly useful fingerprinting study because it moves beyond the usual self-selected visitors to a test page. The dataset covers
24,871,013 Firefox browsers, measured from March 25 to April 17, 2026, and the analysis is broken down by operating system.
A few findings stand out:
- For the most discriminating modeled service, cumulative Firefox protections reduced unique fingerprints from 75.65% to 46.79%. That is a reduction of 28.86 percentage points, or about 38% relative—not the elimination of fingerprinting.
- Results vary enormously by fingerprinter and platform. Baseline uniqueness ranged from about 18% for the modeled Akamai script to 76% for Fingerprint Pro. There is no honest universal “X% of browsers are unique” figure.
- A surprisingly small number of characteristics do most of the work: 80% of Fingerprint Pro’s unique results were already unique after eight of its 50-plus measurements.
- Platform-specific analysis matters. Clamping available screen dimensions reduced macOS uniqueness by roughly 25 percentage points in one model, but by only 0–3 points on the other platforms.
- Simply calculating Shannon entropy for each characteristic is not a reliable guide to its practical fingerprinting value.
The headline number also needs the right interpretation.
A unique fingerprint is not automatically a known identity, nor does it prove that a website is tracking anyone. It means that the tested combination appeared once in this measured population. Persistent tracking additionally depends on the fingerprint remaining stable, the same party observing it again and potentially combining it with IP, account or other information. Conversely, “not unique” does not mean anonymous.
The methodology is strong but bounded: Mozilla modeled scripts from Fingerprint Pro, Signifyd and Akamai against Firefox telemetry. It did not measure every browser population, every fingerprinter or how trackers may subsequently adapt. That distinction between demonstrated capability and an observed tracking incident matters.
For Firefox users, the practical advice is pleasantly boring:
- Keep Firefox current.
- Use its built-in fingerprinting protections rather than piling on exotic spoofing extensions and obscure
about:config changes. A one-off configuration can make a browser more distinctive, and stricter settings can break sites.
- Mozilla says some protections operate in all browsing modes, while additional ones are introduced through Private Browsing and Strict Enhanced Tracking Protection. Strict is a reasonable option if the occasional compatibility exception is acceptable.
- Treat the interactive test as a diagnostic comparison, not a pass/fail anonymity certificate. The companion site says the comparison runs locally and submits nothing.
One versioning wrinkle: the paper and companion site attribute the measured headline protections to Firefox 143, while Mozilla’s current security post says the most recent protections in the broader set shipped in Firefox 145. The work was evidently deployed over several releases; either way, remaining on 143 would miss later improvements.
Sources
