Skip to content
InYourGeek
visiteur@inyourgeek — shell
↹ compléter↑↓ historique⏎ ouvrirhelp
FR

The Monday letter · the tech and security week in one email

Free, every Monday, via Brevo (EU), one-click unsubscribe · your data and rights

Security· 3 min read

uBlock Origin patches a universal XSS flaw in its Firefox extension

A researcher has detailed a uBlock Origin flaw on Firefox that let a malicious site add trusted filters, then run JavaScript on any other site. The fix is out: all that's left is checking your extension is up to date.

Illustration of the uBlock Origin extension in Firefox, with a hidden iframe laid over a web page to hijack a click

On 16 November 2026, a researcher published a detailed write-up of a universal XSS flaw in uBlock Origin, which he says he found earlier this month. According to him, the problem is now fixed and has its own security advisory on GitHub. This article rests on that single account, the one given by the person who found the bug: we have no other source at this stage.

What the flaw allowed

uBlock Origin separates ordinary filters from so-called trusted filters, which are allowed to inject JavaScript into pages. According to the researcher, a malicious site could get the extension to save a trusted filter of its choosing, aimed at a different domain. The injected code then ran in the context of the targeted site, whichever one it happened to be. That is what “universal XSS” means: no site is too important to be on the menu.

The attack needed the victim to click on the booby-trapped page. Once.

The element picker as the way in

The route goes through the element picker, the right-click tool that lets you create a blocking rule without leaving the page. It is displayed in an iframe pointing to an internal extension page, one that can ask the background script to add filters.

The researcher describes four steps:

  • find the address of that internal page, which on Firefox means knowing the random identifier unique to each installation;
  • embed it in an invisible iframe, with opacity set to zero;
  • drive the picker with postMessage messages to drop in the desired filter;
  • position the iframe so that the victim clicks the create button without ever seeing it.

The identifier leaked through a rule that ships in the default configuration. It swaps an advertising request for a harmless script, loaded from a moz-extension:// address that a site could intercept by overriding XMLHttpRequest. The researcher notes that this was not the only possible leak.

The picker normally turns away filters capable of injecting scripts. Yet an option field controlled by the attacker accepted a line break, which was enough to smuggle in a second rule, a trusted one this time. Filters created by the picker are indeed treated as trustworthy, at least until the next time the filters are recompiled.

The root defect, according to the author, is that the picker accepted messages coming from the page. One of the fixes introduces a secret, passed by the content script to the real picker and required from the very first message. In other words, the picker now asks for the password before taking orders from strangers.

Affected versions and updating

The flaw affects uBlock Origin in its MV2 version for Firefox. The researcher says a similar problem, which he considers less serious, also affected the MV3 version, uBO Lite.

The write-up gives no version numbers, neither for the vulnerable releases nor for the fixed one, so we are not going to make any up. It points to the GitHub advisory for those details. In practice, open your browser’s extension manager, check that uBlock Origin or uBO Lite is on its latest version, and force the update if you have automatic updates switched off.

A discovery with models doing the legwork

The researcher says the GPT-5.6 Sol model did most of the research work. He suspected that filter injection could lead to a universal XSS and asked the model to look into that lead, then verified the result himself. Most of the proof-of-concept code was written by Claude. The clickjacking step, for its part, was not implemented.

An ad blocker is, by design, allowed to rewrite the pages you visit. That is the whole point of it, and it is also why “remind me later” is not an answer here.

Sources (1)

Written with AI assistance from the sources cited above, then reviewed and approved before publication by Sébastien Soulier.

The Monday letter

The tech and security week, every Monday in your inbox

What to patch first, data breaches, the articles with pictures. Free, no ads, one-click unsubscribe.

Free, every Monday, via Brevo (EU), one-click unsubscribe · your data and rights