Mozilla’s new research write-up examines browser fingerprinting: recognizing a browser across sites by combining details such as fonts, graphics output and screen size, without a cookie. The paper uses measurements from nearly 25 million Firefox users to identify which exposed details make browsers distinctive.

Mozilla modeled three commercial fingerprinting services against those measurements. Its paper says the tested protections reduced the share of unique fingerprints from 75% to 46% in its evaluation. That is a modeled uniqueness result in a Firefox sample, not a measured fall in tracking across all websites or browsers. The authors work partly at Mozilla, which develops the browser being evaluated.

The defense reduces needless differences in what sites can read. For example, Firefox can hide taskbar dimensions from screen APIs or report a coarse processor count. Mozilla says some protections apply by default; others currently require Private Browsing or Strict Enhanced Tracking Protection. Users who want the extra protections in ordinary windows can select Strict in Firefox’s tracking-protection settings, then check individual sites if functionality breaks.

What is newly published

The paper’s extended version is dated October 6, 2026; Mozilla’s security blog post was published November 10. The underlying features are older. The paper identifies Firefox 143 for part of the deployment, while Mozilla’s Firefox 145 announcement is dated November 10, 2025. This is new public evidence and explanation of deployed defenses, not a newly issued browser patch.

What the thread adds

In the Hacker News discussion, rdmuser notices that the linked Firefox 145 launch post is from last year. Mozilla’s first-party dates confirm that correction to the timeline. mitxela replies to a proposal to block fingerprint data from leaving the browser. Layout choices based on screen size could still be inferred from user interactions, making such blocking difficult without breaking sites. That is a reader’s technical objection, not a result measured in Mozilla’s study.

From a 53-comment thread, in HN’s own order.

Caveat: simulated fingerprinting services and Firefox-only telemetry cannot tell us how often real trackers identify a person over time.