Universal XSS in uBlock Origin from website-driven element picker

594 points by zaik 21 hours ago on hackernews | 189 comments

ByteOfWood | 21 hours ago

I never really loved how extensions have to put their interfaces in the website's DOM if they want to be able to work with the page's elements, like this selector, or a password manager's suggested logins. Especially in the password manager's scenario, it could leak what accounts you have to a given website to that website itself, if the page is looking for new DOM elements. Not exactly a showstopper vulnerability, but it's not a good thing.

esprehn | 17 hours ago

I've always regretted we didn't push harder on an extension overlay API. It was supposed to be based on Top Layer, but the primitive version of that API was never added and the things that were added can't be effectively used to attach subtrees for extensions.

https://developer.mozilla.org/en-US/docs/Glossary/Top_layer

As a fun fact the Devtools also have a separate "Overlay Page" to draw on top of the web page for specifically this reason. The date picker on some platforms similarly uses an overlay page. This machinery was just never exposed to extensions.

btown | 20 hours ago

Setting aside data leaks, this also causes bugs in extensions to potentially affect websites en masse. For instance, as recently as this January, 1Password began injecting prism.js globally on all pages, breaking many <code> blocks for at least 20 days: https://mjtsai.com/blog/2026/01/07/1password-browser-extensi...

dawnerd | 17 hours ago

Speaking of 1pass, their overall software quality has taken a massive nosedive. I’m almost at the point of ditching them.

Same. User of 15 years. Solid native apps turned into buggy Electron mess. I'd be fine with Electron if it weren't buggy. Wi-Fi server went away. 25% price increase this year and they do stupid sports sponsorships now.

malfist | 11 hours ago

I wonder how much of this is to be blamed on ai psychosis C-suites have. Everybody I know has been complaining that the push from the c suite down is to ship ship ship with AI. Proper engineering is seen as a waste of time, along with testing and future thinking. Companies just ship more and more untested slop

ceejayoz | 10 hours ago

They took $200M outside investment in 2019.

https://1password.com/blog/accel-partnership

Enshittification began shortly thereafter.

malfist | 10 hours ago

Right, but what part of VC funding means ditching native apps for electron? Enshitification usually means exploitation, this is just change for change's sake

wlesieutre | 9 hours ago

Prior to that the Windows version was a somewhat half-baked thing made by a contractor and didn't keep parity with the Mac version. It was better than nothing if you had to use Windows sometimes, but you wouldn't have bought it just for that version.

But with the VC funding came a change from B2C sales to Mac users who wanted good software to targeting B2B enterprise customers who wanted a feature checklist across their fleet, Windows included.

ceejayoz | 10 hours ago

Ditching native apps means ditching native developers?

Fordec | 15 hours ago

And dump money into Omarchy

swozey | 17 hours ago

Took me 10 minutes to set up vaultwarden in a docker container and migrate all my 1p to it last time they pissed me off earlier this year

gib444 | 16 hours ago

Genuine question: is it actually better, or have you not discovered its own issues? Ie did you did a decent comparison and analysis before migrating?

Cockbrand | 15 hours ago

Not GP, but I did the same about 3 years ago when 1PW transitioned from 7 to 8: it works decently, but the UI/UX is not as polished compared to how 1PW was back then. No technical issues at all, just not as smooth. It's still fine and not something I notice any more.

noname120 | 15 hours ago

I did this, the UX of Bitwarden’s UI is horrific and I switched back after holding for a year.

dust-jacket | 12 hours ago

I find this kind of claim baffling.

As someone who uses one for work and the other for personal pws, I'm puzzled about what you're doing in the UI of either that makes BW even all that different, let alone 'horrific'

QuantumNomad_ | 7 hours ago

1Password does make it much nicer to organize all kinds of passwords and keys in a way that nothing else I’ve tried even comes near to.

Especially the ability to create sorts of documents that contain as many custom sections as you want. At the top I put an intro text to explain to myself a bit of context and then I have fields with for example paths to files like ssh keys and for each of those a field with the password for that ssh key.

I also store software license keys in documents in 1Password.

And I have documents that contain cli commands for some operations that require different passwords each in order followed by the corresponding passwords, and interleaved with longer sections of text providing explanations and context.

weaksauce | 3 hours ago

the ux of bitwarden is ok but very spartan. it certainly works but there’s no extra thought added to anything to delight the user. one small example is clicking on a folder doesn’t open the folder… you have to click the tiny arrow. this is akin to a form input on the web that doesn’t take the time to use labels without a form for attribute providing a similar affordance. small thing but annoying enough when you look at it in totality.

cstrahan | 7 hours ago

I use BitWarden, and I share the disappointment in the browser extension UI/UX.

Too many clicks to do something trivial. Have to click the tiny hamburger icon to the right of an entry to get to “View” or “Edit”. If I’m viewing an entry but then want to edit, I have to mouse over diagonally to click the “Edit” button in the bottom left corner.

The View vs Edit distinction is pointless friction. Let me edit the fields directly. Or at least make it less tedious.

Clicking the entry itself should take you to Edit, not perform the fill password action. If I wanted to fill, I can do that with a keyboard shortcut or via the overlay when a password field has focus. Thus, edit is the most common reason I’d want to interact with an entry in the extension pop up. But I have to precisely mouse over to click the hamburger icon to next click Edit/View.

These are “small things”, but then again a paper cut is a “small thing” too — doesn’t mean I’d be happy to have a dozen paper cuts every day.

thedanbob | 3 hours ago

> Clicking the entry itself should take you to Edit, not perform the fill password action.

This was probably the single most annoying change when they redesigned the extension. Apparently they're bringing it back soon, at least as an option.

swozey | 10 hours ago

It's definitely not better, 1p is by far the best manager, but it works fine. I still have 1p installed because I was paying yearly and I'm running bitwarden in a container that sometimes I forget to restart, etc so its more manual, but I don't think 1passwors was worth the yearly fee before their Omarchy donations etc, and I don't even want to think about how much money I've paid to 1password for storing a 3mb database for me for 10 years.

I had been wanting to get my passwords off the cloud for 5 years and just finally did it. Everything transferred fine, but sometimes a UX or chrome extension glitch happens and I'll wind up storing a new pass in 1p instead of bitwarden. Before my subscription ends in 6 months I'll just dump 1p again and import it and fully delete the app once I can't log in again.

You just aren't getting that super smooth 1p UX. It's more like a linux app that has quirks, weird design decisions, etc. Like its 2 clicks to copy password, not just clicking "copy password" like in 1password. And after 10+ years of 1p UX over and over every day it's a bit jarring but its a LOT better than lastpass was 10 years ago and totally fine. It, generally, looks and acts almost exactly like 1p.

The container is super simple, I just start it and open bitwarden. I have never had to troubleshoot it and it starts almost instantly and is ready to use.

Use tailscale and put it on a magic dns name like "vault.mytailscale.net" and use it everywhere you have connections.

def_true_false | 9 hours ago

The Bitwarden extension has a setting that shows a separate copy username and copy password buttons, instead of the menu that contains both.

swozey | 8 hours ago

My chrome extension for bitwarden is probably the buggiest thing, but I just upgraded to Golden Gate and that broke my touchID so it may be related. HAlf the time I click it it doesnt open the UI at all

techpression | 16 hours ago

Same here, it’s sad how they went from working flawlessly with a quick and clean client, to whatever it is today. Sometimes I get three or more popups when trying sign in (many after the login has already been completed)…

drcongo | 13 hours ago

Same. My favourite is the one offering to save the login that you've just filled from a saved login while simultaneously using 12GB of RAM. What an absolute clown car 1Password has become.

benhurmarcel | 13 hours ago

1password importing a third-party script on all webpages also had me chasing a CSP policy console error on a web app I was making.

saagarjha | 15 hours ago

I’m very confused why 1Password should be doing syntax highlighting for code at all…

brycewray | 7 hours ago

As I recall reading at the time this occurred, the additional Prism.js code apparently was mistakenly left behind from a particular inter-version development test. This was fixed in a subsequent 1Password update. The first release notes mentioning the fix[0] were for v.8.12.0:

> We’ve fixed an issue where the 1Password extension could break syntax highlighting for code blocks on some websites.

[0]: https://releases.1password.com/b5x/stable/#1password-in-the-...

jareklupinski | 7 hours ago

> 1Password began injecting prism.js globally on all pages

https://en.wikipedia.org/wiki/PRISM

> PRISM is a code name for a program under which the United States National Security Agency (NSA) collects internet communications from various U.S. internet companies.

people are really bad at naming things XD

wtallis | 20 hours ago

Did XUL extensions in Firefox prior to WebExtensions have a different way of interacting with the page contents? I know they tended to integrate better with the host browser's UI and inherited more of the host OS's UI conventions as a result, but the web back then didn't require as many CSS-based ad blocking rules (domain/URL-based rules worked for almost everything) so I don't remember what the element picker UI was like on eg. AdBlock Plus or Adblock 0.5.

forgotmypw17 | 20 hours ago

XUL extensions basically have full access to the entire browser's UI and internals, which allows for some very powerful extensions.

If this sounds appealing to you, you may be happy to know that XUL lives on as Unified XUL Platform (UXP) and is still supported by currently maintained browsers.

UXP: https://repo.palemoon.org/MoonchildProductions/UXP

Pale Moon: https://www.palemoon.org/

Basilisk: https://www.basilisk-browser.org/faq.html

SeaMonkey: https://www.seamonkey-project.org/

wtallis | 19 hours ago

That doesn't really answer the question. Sure, XUL extensions have better access to the internals of the browser. But does that translate into XUL extensions being able to better isolate their own internals from being accessed by untrusted websites?

rafram | 10 hours ago

No, the opposite. Firefox WebExtensions inherit the two main security features that XUL extensions used to separate page content from extension code. But in the XUL extension days, if page content did manage to control an extension, it could directly take over your machine and execute any code with full filesystem access.

(The two security features are X-ray vision and cross-origin access controls. The former makes it so extension code doesn’t see the page’s modifications to built-in properties. For example, if my extension calls document.createRange(), but the page has overwritten that method, my code will see the original and not the modified version. But if I call document.wrappedJSObject.createRange(), I ask to get the page’s version, which might be malicious. The latter just means that objects created by extensions and made accessible to the page can’t actually be used by it unless the extension explicitly exports them. Both of these are still active for WebExtensions.)

Lotta rose-tinted glasses when people remember the XUL extension days. It was an extraordinarily risky and fragile architecture. Every single extension had full control over your machine, and every major Firefox update broke most extensions (or made them break Firefox), because the main way to inject your logic into the browser was monkey-patching.

forgotmypw17 | 15 hours ago

My intuition was yes, but research says otherwise. They have had their own vulnerabilities. Basically, anywhere anything interacts with site code, there's potential for vulnerabilities. It might be better, but the extensions still read HTML from the page, and... guess what can happen then?

forgotmypw17 | 19 hours ago

Yes, with caveats. They could have UI that the websites couldn't access, but the fact that they interacted with website content could still expose them to attacks.

creatonez | 8 hours ago

Pale Moon is backporting theater, I would highly recommend against using a browser that is jam-packed with so many bugs.

mimasama | 10 hours ago

At the very least the author's method doesn't seem to work for the "legacy" version of uBlock Origin here in Pale Moon since the iframe uBO adds to the DOM is an HTMLDocument pointing to about:blank. There are also no moz-extension:// shenanigans since uBO lives in chrome:// which can't be accessed from content I believe

XUL extensions could essentially open pop-up windows (the way context menus do), so yes. It can be fully transparent if needed, I believe.

You can probably drop an absolutely positioned block (relative to the browser chrome, outside the tabbrowser) too, but I think that didn't work in some older versions?

Note that this is all based on very vague memories, so may be off… Also, both scenarios don't bother injecting anything inside the content. This unfortunately means you need to do some annoying math to position things correctly (in particular if funny DPIs are involved).

flomo | 16 hours ago

Yes. There was even an XUL extension which loaded ActiveX controls, except without IE's malfunctional 'security zones'. Honestly, that was always a massive security blindspot.

This leak doesn’t exist. The injection goes in a sandboxed iframe, which is protected from the rest of the page by the same security boundary that prevents them from puppeteering your bank’s pages in an iframe they’ve embedded themselves.

Banou | 19 hours ago

In the case of password managers, amy decent one will put their DOM elements in a closed shadowDOM root, which prevents the page scripts from interacting with it entirely.

forgotmypw17 | 19 hours ago

Shadow DOM is not a security boundary, and can be bypassed.

Banou | 16 hours ago

I didn't go into full details but it's usually a combination of shadowDOM plus Iframe for the security boundary

ByteOfWood | 5 hours ago

This is correct, I just checked with Bitwarden. They use an iframe with the extension's src, nested within a closed shadow DOM. I was not able to get any data from the extension's iframe, although the site can still detect the specific extension is in use which could be used for fingerprinting.

moi2388 | 19 hours ago

From a page, which loads after the extension? How?

forgotmypw17 | 19 hours ago

>The {mode: "open"} argument gives the page a way to break the encapsulation of your shadow DOM. If you don't want to give the page this ability, pass {mode: "closed"} instead, and then shadowRoot returns null.

>However, you should not consider this a strong security mechanism, because there are ways it can be evaded, for example by browser extensions running in the page. It's more of an indication that the page should not access the internals of your shadow DOM tree.

https://developer.mozilla.org/en-US/docs/Web/API/Web_compone...

This presentation talks about using window.find() and other methods:

https://speakerdeck.com/masatokinugawa/shadow-dom-and-securi...

moi2388 | 7 hours ago

1. Correct. Open is peekable, closed is not.

2. Extensions can do whatever the user lets them. A web page however can not access a content script closed shadow root

3. Prototype pollution also cannot occur since content scripts run before the page

4. Window.find() also does not work if you run the content script in an isolated world, as you should.

If you do all of the above, a web page cannot access data from your content script closed shadow root.

Browser extensions can. Running untrusted browser extensions is extremely dangerous.

WhyNotHugo | 9 hours ago

A password manager should prompt the user to pick which extension to share, and send only that to the browser extension itself.

I'm aware that several show all accounts in the DOM itself _and then_ prompt to disclosed only the secret material, but that's honestly not great design.

mattashii | 12 hours ago

> have to put their interfaces in the website's DOM if they want to be able to work with the page's elements

Can't the extension use the sidebar? I know Tree Style Tab uses it, so why couldn't an extension put the selector there?

taylorfinley | 21 hours ago

Really love the "I wrote all the prose, gpt wrote the POC" disclaimer, I would love to see more of those kinds of signals on human authored content.

qn8bdebq | 21 hours ago

Instant vote of confidence, yes!

MOARDONGZPLZ | 20 hours ago

I wouldn’t say that, but it’s honest. And I’ll read the write up and reserve judgement until I’m done or decide it’s not worth continuing. When AI writes the article and there’s a disclaimer, I won’t read it. Saw one the other day: I used ai to write this, if you have a problem with that, stop reading. Of course, I did stop reading (and flagged the post).

kbenson | an hour ago

> Of course, I did stop reading (and flagged the post).

Why would you flag the post? Generated content in submitted links does not break the guidelines, just AI in content of submissions (e.g. Ask HN) and comments. It's not definitionally inappropriate or of no use either, unless posting to a website that auto-generates documentation out of some other content (e.g. module documentation or man pages) would be. For example, where it provides relations and links and dependency information by scanning the code, which often has problems where it determines a soft or optional dependency is a hard one.

satvikpendem | 6 hours ago

Not sure why your opinion was flagged, some people don't want to read AI written content as you mentioned.

qntmfred | 3 hours ago

some people don't want to read people written AI content disclaimers

jumperabg | 18 hours ago

This might be the last vulnerability of the year with human prose in it.

hiccuphippo | 19 hours ago

If there's enough of them and AI trains on it, then AI will write that disclaimer itself to make itself more believable.

techpression | 16 hours ago

Only if it’s part of a benchmark that matters, which is the reason we are where we are today, good writing is not benchmarked enough.

10111001 | 10 hours ago

> Only if it’s part of a benchmark that matters, which is the reason we are where we are today, good writing is not benchmarked enough.

It's that the problem? "Good writing" is benchmarked, but we get writing the over-fits the benchmark?

charcircuit | 19 hours ago

No, it won't. That is not how things work after all the fine tuning and alignment that gets done.

denkmoon | 17 hours ago

Why not? Genuinely not seeing it.

charcircuit | 13 hours ago

Imagine trying to poison datasets to make ChatGPT respond that it is James Bond when a user asks "Who are you?" I personally can't see how that would be possible to do. OpenAI explicitly trains it so that it knows it is ChatGPT, they provide it in the system prompt, and it would be caught before they released the model if it did somehow happen.

praash | 17 hours ago

Fine tuning and alignment has not prevented other kinds of hallucinations either.

charcircuit | 13 hours ago

But it has. Typically people have to do some kind of jailbreak or ask in a funny manner to get it to go around its training.

hiccuphippo | 7 hours ago

People already ask it to write in a natural way that doesn't look like AI, wouldn't adding a disclaimer that it's not AI the most effective way to not look AI? I've seen people do a 180 from dissing something for looking AI to praising it when they learn it was not, even when the claims are unverifiable. Wouldn't AI learn that and use it?

charcircuit | 47 minutes ago

That is out of scope. Of course people could explicitly prompt for such a section to be added to pretend not to be AI, but that is not what we are talking about here which is prompting for a summarization and the AI itself deciding to pretend to be a human.

shevy-java | 18 hours ago

Of course AI could.

jethro_tell | 18 hours ago

Uh, ok

nvme0n1p1 | 18 hours ago

Alignment is solved? Fantastic news! Pray tell, where can I read more?

charcircuit | 13 hours ago

For simple problems yes. Look up the term "blue teaming." If researchers ask AI to generate a summary of a vulnerability and then it randomly ads a disclaimer that is something that can be trained out of the model.

nvme0n1p1 | 8 hours ago

You seem misguided. A blue team doesn't prove the non-existence of vulnerabilities. It's just a way to try to get ahead in the cat-and-mouse game with attackers.

This is like saying "car crashes are solved. look up the term "airbags"

charcircuit | 48 minutes ago

Except this isn't really a cat and mouse game. Essentially we are covering a prompt "Summarize this security vulnerability" and ensuring it doesn't claim to be human in the summary. There is no mouse here like trying alternate prompts until it happens. We are talking about basic prompts where we see bad behavior. Basic enough prompts for a blue team to check.

Ensuring the safety of passengers can be stated to be "solved" for a specific controlled case.

paganel | 16 hours ago

> after all the fine tuning and alignment that gets done.

This does indeed sound like religious belief.

charcircuit | 13 hours ago

I'm simply pointing out that modern LLMs do not just regurgitate what was most common in the dataset for the base model. Especially when we are talking about how it is structuring the output of a generic task it's doing.

lucumo | 16 hours ago

If it works, we'll see people lying about it long before AI trains on it.

jonhohle | 10 hours ago

UBlock (and possily many more) leaks its UUID? So by installing an ad blocker, you’ve just given websites a unique fingerprint to reliably track you?

planteur | 9 hours ago

Related: “LinkedIn checks for 2953 browser extensions” https://news.ycombinator.com/item?id=46904361

deepsun | 6 hours ago

In Chrome only. Firefox doesn't allow that.

briaoeuidhtns | 6 hours ago

they may not allow directly checking extensions, but I'm almost positive the trick in the article still works

buckle8017 | 5 hours ago

it doesn't, Firefox assigns random ids to each install

But that's even better!

You can track by any of these IDs rather than by their combination.

buckle8017 | 38 minutes ago

You can't enumerate them.

It's (supposed to be) a secret.

0rzech | a minute ago

But because of how Firefox extension API works, its extensions may leak their unique installation IDs to visited websites. It's much, much worse and basically nullifies any other privacy improvement in Firefox. Such UUID is more than enough to successfully track specific person.

forgotmypw17 | 20 hours ago

For those checking (I still had the vulnerable version), you should be at these versions or higher:

uBlock Origin: 1.75.4

uBO Lite: 2026.1110.1816

forgotmypw17 | 19 hours ago

Thank you, I corrected my comment.

olibhel | 19 hours ago

After the update, I still have 1.76.2 on the firefox dev edition on mac.

AlexAplin | 19 hours ago

The advisory page says 1.75.2 and lower, patched in 1.75.4: https://github.com/gorhill/uBlock/security/advisories/GHSA-c...

spopejoy | 19 hours ago

This happened to me. Apparently you have to update twice to get to 1.76.4.

cubefox | 13 hours ago

Apparently it's not possible to manually update extensions on Firefox for Android.

norenh | 7 hours ago

After quite a lot of trying, I came to the same conclusion. However, removing and adding it again does work which will give you the latest version.

cubefox | 7 hours ago

That might remove custom settings though.

akersten | 20 hours ago

So this seems totally solvable by the create-rule element living in the browser chrome instead of the website DOM... why aren't we doing that?

mimasama | 10 hours ago

It used to be. The "legacy" version of uBO intended for Pale Moon and pre-Quantum Firefox had the add-on living in chrome://ublock0 (the picker is in chrome://ublock0/content/epicker.html), and the iframe for the element picker injected into a website's DOM is an HTMLDocument pointing to about:blank.

user070223 | 15 hours ago

I've 3rd party frames blocked by default, which is good reminder to everyone that you can up one your blocking game with increasing the blocking mode level as described here https://github.com/gorhill/uBlock/wiki/Blocking-mode.

One can have keyboard shortcut which automatically reduces that blocking level(may cause refresh) if something not right, or enable on domain basis. I got used to enabling stuff when have issues and it's not mentally taxing

its-summertime | 12 hours ago

I'm not to test it, however, would ublock consider it a 3rd party frame, especially when it is a frame src that ublock itself injects into pages?

mac-attack | 7 hours ago

Same. There's friction initially, but most users visit the same sites regardless, so the effort drops dramatically over time.

As an added benefit, you unconsciously get a feel for what connections are useful for the user versus useful for the advertiser and can make global blocks/allows as you see fit.

abcd_f | 10 hours ago

Hoooold on ...

> the extension internal UUID which is random per-installation

... and it's possible for a loaded page to figure it out. It sure sounds like this gives us a stable unique tracking ID ... ?

afavour | 21 hours ago

I understand why folks railed against web extension lock downs that blocked directly editing page content but this is a great example of why it’s a powder keg.

contravariant | 14 hours ago

The other way round it looks like a website's javascript being allowed to interfere with extensions is also a bit of a problem. I can't really think of a reason you'd want to let it interecept requests to a moz-extension: URL or allow it to make an iframe for one.

Neither could be considered a security bug, but it's a bit odd the entire interface is wide open.

Maxion | 17 hours ago

Malicious ads are a whole 'nother attack vector that is probably a larger issue.

whywhywhywhy | 13 hours ago

The issue isn't modifying the page, the issue is the page is able to interact with the modifier.

uptrails | 8 hours ago

The concern was much less about vulns and a lot more about the pattern of monetizing free extensions by siphoning off browsing history, injecting ads, etc. This was very common; even some "privacy" extensions would do that.

Another worry was the possibility of extensions being hijacked or sold and then doing something outright malicious, e.g. stealing passwords / cryptocurrency.

AnthonyMouse | 16 hours ago

Chrome itself can let go of multiple remote code execution CVEs in the same month, but an extension bug that allows XSS is a powder keg?

GCUMstlyHarmls | 20 hours ago

Wonder what the over-under is on a page exploiting your `block-element` interface if you happen to open it (I guess you could sort of encourage that interaction) vs explicitly getting served malware eg "Discover-Tech’s Controlium".

badsectoracula | 14 hours ago

Uh no it isn't, being able to edit the page contents is the entire purpose of this extension (and many others), the issue here is that because there is no other way for an extension to provide a UI to the user it also has to do that by modifying the page - which is something a malicious page can take advantage of.

The solution here isn't "extensions shouldn't be allowed to modify the page" but "extensions should have some means of creating their own UI so they don't have to modify the page for their UI".

alex1138 | 18 hours ago

A la Chrome/manifest 3? Not sure if we're on the same page here or if you mean something else

Ikersl | 7 hours ago

its nuts that an ad blocker could have such vulnerabilities

I recommend to delete old forefox profiles and create new once, than install new fixed uBlock0. You should always use separate profiles for all tasks with hardened user.js anyway. Don’t forget to save your bookmarks. Anynew profile creates new UUID.

bita_nidir | 13 hours ago

Guys, go to about:addons > cog > check for updates, and update your uBlock Origin. Mine had not auto updated.

renehsz | 17 hours ago

Is AdNauseam (a fork of uBlock Origin) also affected?

cdman | 16 hours ago

I had the same question. This commit: https://github.com/dhowe/AdNauseam/commit/0e757ee3d1fd1b5406... seems to be similar to the one fixing the issue with uBlock Origin: https://github.com/gorhill/uBlock/commit/bf69cf7405d7d9e7e4b... (look at the changes in src/js/resources/utils.js), so I think the answer is "yes, it's affected" and also "yes, it's fixed in the latest release".

creatonez | 3 hours ago

AdNauseum isn't exactly a hardened option in the first place. Running the ads in the background instead of nuking them out of existence is dubious and seems like it would be complex to get right.

db48x | 16 hours ago

Assume that it is until you have read through the source and proven otherwise.

SoftTalker | 21 hours ago

uBlock Origin has broad permission to inspect and modify anything on a page (it needs it, to do what it does). This would seem to make any vulnerability easier to exploit.

hackerbone | 14 hours ago

Browsers should implement something like a sandboxed DOM where execution is limited to a snapshot of the DOM. This would prevent most of the client side related vulnerabilities.

jimmydorry | 13 hours ago

That sounds like a halving of performance for marginal security gain, when we have such glaring holes like extensions having unfettered access to all network requests made by the browser (the gap that manifest v3 moved towards closing).

chmod775 | 12 hours ago

The entire point of browser extensions is to operate outside the default sandbox. They're browser extensions, not browser content extensions.

On one hand we're widening what a tab may do, throwing dozens of permission prompts at users until they're fatigued, on the other we're trying to lock down extensions to the point of making them redundant. I wouldn't be surprised if by the time manifest v5 rolls around, tabs and extensions will be indistinguishable.

At that point someone will create a chrome mod "SuperExtensions", sourcing extensions from a third-party marketplace that is worse and less secure in every way than what we had in 2020, but at least gives people control over the software that runs on their hardware with actually useful exensions again. Such as adblock that doesn't suck. Or an extension that neuters spyware built into browsers.

But of course it wasn't about those extensions at all. It was all about security, not Alphabet's bottom line. Because we can trust people to vote, drive, drink, and do plenty of things, but we can't trust them to decide what software runs on their computer. The paternalistic security argument of the technological infant. Honi soit qui mal y pense.

Edit: Sorry for the only tangentially related rant.

fg137 | 11 hours ago

You see, the company behind the most used browser isn't motivated by this.

tasuki | 14 hours ago

> GPT-5.6 Sol did most of the hard work of finding the issue

I wonder why 5.6 rather than 6.1 ?

wontforgetdisda | 12 hours ago

Either they started long ago or used solely free Copilot..

Multicomp | 11 hours ago

or they, like me, found 6.1 to be a regression in capabilities compared to 5.6. 5.6 Luna likewise is a bigger model than 6.1 luna

possible workaround

      * * 3p-frame block
Dynamic filtering: Benefits of blocking 3rd party iframe tags (by default)

* https://github.com/gorhill/uBlock/wiki/Dynamic-filtering:-Be...

anthk | 13 hours ago

I prefer the builtin Rust based (more performing) ad blocking engine from Waterfox and on top of these, instead of UBo, I uset NoScript. The most secure code it's the one which doesn't run.

stratos123 | 11 hours ago

NoScript breaks most websites, and while I'm willing to believe WaterFox's impl is more performant, it probably has just as many or more hidden security problems as UBlock, since fewer eyes have looked at it.

ethersteeds | 21 hours ago

> But the attacker website can itself hook the XHR made from inside the no-fetch-if hook to extract the random extension UUID.

Oh great. I recently learned[0] that some Firefox extensions can be tricked into leaking their install-unique UUID, which can then be used as a tracking identifier. Today I learn that any site can get my unique uBO UUID by hooking fetch('/adsbygoogle.js').

That is... an ironic disaster.

[0] https://news.ycombinator.com/item?id=50454639

tech234a | 20 hours ago

Chromium used static IDs for web accessible resources until very recently, and even now it's only optional for extensions to use dynamic IDs. This page enumerates extensions with non-dynamic URLs in Chromium: https://browserleaks.com/chrome

esseph | 18 hours ago

I found that by having "Do Not Track" enabled in my browser, I went from nearly untrackable, to within 0.12% of the population.

Ironic.

RetroTechie | 11 hours ago

"I found"

How?

esseph | 9 hours ago

asimovDev | 16 hours ago

I like how the latest commit on uBO Lite's repository is "create SECURITY.md" https://github.com/uBlockOrigin/uBOL-home/commit/e01ffb036d9...

no advisory still on the github page unlike the big sibling's mv2 repo

maplet | 20 hours ago

Good day to use EWW

burgerone | 17 hours ago

I'm thoroughly disappointed that a vulnerability as trivial as this requires AI assistance. The poster already had all the important information at hand.

charlie2agent | 12 hours ago

[flagged]

yoy0721 | 12 hours ago

[flagged]

spencerawlson | 9 hours ago

[flagged]

lakhya | 19 hours ago

[flagged]

[flagged]

bzvox | 15 hours ago

[dead]

usr1106 | 20 hours ago

Haven't used any uBlock for many years. Use DNS based blocking.

For many years I was a happy user of CAD (Cookie Auto Delete). Unfortunately it seems unmaintained. I have now stopped to install it. Don't understand enough of modern browsers to even have an educated guess what kind of vulnerabilities could be buried in the old code.

etherlink | 19 hours ago

DNS-based ad-blocking is very ineffective in the modern web browsing scenarios. You might block most ads from your local news site, although YouTube, Twitch, Facebook and any other major site has more sophisticated ad-networks which require ad-blocking at the browser level.

You've been required to use a browser based ad-blocked on YouTube for over 5 years now.

usr1106 | 19 hours ago

I have not used Facebook for many years. My Youtube usage is well less than an hour per month. Those are shitty businesses I don't want to waste my time with. Generally my Web usage is going down. I hate most of the internet these days.

72deluxe | 12 hours ago

I think being very conscious in what you are looking at and reading or using the Internet mainly for research like we used to do back in the day (Geocities obscure sites!) is probably the solution. The alternative is GIGO

nvme0n1p1 | 18 hours ago

It's great that you don't use the internet. We're all happy for you. Other people do use the internet, and they shouldn't get their advice on how to use the internet from someone who doesn't use the internet.

cheschire | 14 hours ago

SponsorBlock, set to allowed on YouTube only.

> Haven't used any uBlock for many years. Use DNS based blocking.

You're missing out (or rather, getting a lot more ads and trackers and spyware).

DNS blocking is complementary, and a good thing to also have. Some sites are best blocked at that level becuase they are pure evil.

But DNS blocking will miss most of the spyware/adware/malware that uBlock Origin can handle.

DarkmSparks | 14 hours ago

so this is the security infrastructure manifestv3 created?

I'm no expert but that doesn't seem very secure to me, maybe there was alterior motives.

negura | 19 hours ago

The concept of browser extensions seems terrible in general. There are many that ship with malware, botnets, infostealers etc. uBlock Origin even requires a compiled .wasm module. It's unwise to grant that level of complexity from third party devs permission over something as sensitive as your web browser.

ada-h00k | 12 hours ago

The solution would be to properly implement the features that enough users want into the browser. After all, extensions are usually just a stopgap measure. Firefox could also simply build in an ad blocker that works just as well as uBlock Origin. For password managers, they could simply build an API into the browser that password managers could then use to integrate with the browser. Then there would be no need for an extra extension. The same goes for things like add-ons to download videos from sites that prevent it, such as YouTube or similar streaming sites. If Firefox simply had a built-in feature to scrape video files from websites and save them normally via the right-click menu, no one would install dubious extensions for that purpose.

TeMPOraL | 11 hours ago

That's an argument in favor of extensions. Most of the examples you listed are features that oppose the interests of most commercial players on the market, including but not limited to, many multinational corporations. If they were all first-party features of the browser, Firefox would've been dead a long time ago.

As it is, they're just voluntary extra software that Firefox neither bundles nor recommends explicitly, which keeps both the browsers and extensions safe - merely supporting arbitrary extensions in general is not enough to create focused market pushback against the browser (and gives Fx some leverage, mostly around "accessibility"), while individual extensions are too small a target to fight.

It's a good separation of concerns, and a somewhat stable equilibrium.

ada-h00k | 11 hours ago

A browser should not be concerned with the interests of commercial players. Software should, first and foremost, take the interests of users into account. If they try to do something against Firefox, we have to fight back. Just going along with it and giving in will only lead, in the long run, to the Internet becoming even more broken than it already is.

canicollectongo | 19 hours ago

Th day no browser allows it is the day I make the vibe-browser. A browser that scraps all the content of every website, and tells an AI to present it without the garbage.

This extensions exist to solve a problem where the alternative will NOT involve playing nice about it.

numpad0 | 15 hours ago

Start it today... The best time to plant that tree may have been 15 years ago, you may still have 5 years for it to be viable if you start now. There no need to wait until the quote matures to start panicking.

Telaneo | 19 hours ago

The same can be said about websites themselves.

madeofpalk | 19 hours ago

Well no not really. There’s strong cross-origin restrictions in browsers to prevent anything bad happening just because you visit a website.

Web extensions are troubling because they tend to have pretty unfettered access to one of the most sensitive parts of your computer.

AnthonyMouse | 16 hours ago

We're talking about bugs here. If the website has a bug you're not just compromising one user. Input sanitizing bug, or the site even just makes the mistake of hosting user-uploaded content on the same domain as the main site, and the attacker can upload javascript that then gets run under the origin of the main site. Attacker can now capture other users' passwords or otherwise take control over their accounts because the user visited the main site with the bug rather than needing to visit the attacker's. Many of those passwords will have been reused on other sites.

Meanwhile an extension can't do anything the browser can't do, so why is it a problem for the extension but not the browser itself?

negura | 19 hours ago

Individual websites can't be granted permission over your entire browser, they operate in a sandbox (then again firefox for example has a relatively insecure sandbox).

account42 | 14 hours ago

Yes we should only grant that level of access to hostile advertising megacorporations.

TZubiri | 21 hours ago

Software professionals should uninstall a javascript plugin that runs on every website whose purpose is to consume content more comfortably.

It’s not about comfort. It’s basic security at this point. 2026 web ads are cancer.

cdman | 16 hours ago

^^^ This. As someone who worked in the computer security for years: the second best thing for security (after turning on automatic updates for your OS / software that you use) is to block ads. They are chock full with scams in addition to plain old malware.

LoganDark | 16 hours ago

That depends on the OS / software in question.

jaharios | 12 hours ago

ButlerianJihad | 17 hours ago

I used uBlock Origin and I enjoyed the relief, I must say. But, it did slightly nag at me that I was "ripping off resources" from the very sites that I should've been supporting in that way. I mean, if I go and visit every day and get some benefit out of them, their webmaster team deserves a cut of the proceeds, at least, yeah? But I am a freeloader by nature and unfortunately, the Internet spawned a generation of freeloaders. Not that there's anything wrong with freeloading. I've always supported RMS and the F/OSS vision. (Those licenses are not freeload licenses!)

Anyway, Edge is ramping up some decent protection against tracking and ad-annoyances. Again, I'm posting on the wrong thread, wrong site about these alternatives. But I've found Edge already blocking stuff that was blasting through my Chrome sessions.

And then there's Chrome, which offers two levels of Safe Browsing protection, as of November 2026. Have we enabled the highest AI-heuristic-powered levels of safety for yourself? Are we at high risk of malware attacks because we're klicking on sketchie linkies all day? Yeah... but we're too paranoid to touch Google kit anyway? Google also has the Advanced Protection Program we can enroll in, if we're a palestinian reporter or Elon Musk type.

Just another reminder that many don't think about: if we are adding extensions to our browser like there's no tomorrow, all of them are increasing our attack surface. (UBO excepted...) These risky integrations and extensions are a really fertile field for malware, spyware, and supply-chain attacks. We hate Manifest v3, but we're running 2 dozen extensions, any one of which could stab us in the back next month, next year, next millennium falcon.

tl;dr: Thank you kindly for the downvotes

zelphirkalt | 16 hours ago

You'd trust Google to protect you, if you were a Palestinian journalist? Haha, tell me you are joking, please.

Supporting RMS and feeling like blocking ads is "ripping off resources" from websites? How did you make those 2 things square up in your mind? They are like opposites! And then more of the same, talking about MS Edge. Also talking about Google Chrome and not at least Chromium, or better Ungoogled Chromium.

And why does one need to "enroll" in something something Google anyways, in order to get what should be in the default package?

It all sounds very unreflected.

jeffbee | 20 hours ago

[flagged]

mahboi | 19 hours ago

I would never use a browser with any extensions for anything important. Firefox with uBO is there for casual browsing. MV3 is useless for ad-blocking.

LiamPowell | 18 hours ago

What universe are you people living in? Every ad blocker except uBO has been using DNR for years and they all work fine. uBO Lite uses DNR and it works fine (although recently it stopped being permissionless, which I think was a huge mistake).

Every time MV3 comes up there's a slew of people that say you can't do ad blocking with it but no one ever prevents evidence or explains why everyone not using uBO has working ad blocking.

> I would never use a browser with any extensions for anything important.

I agree with this though, or at least I agree for extensions that ask for overly broad permissions such as uBO and uBO Lite.

syradar | 17 hours ago

MV3 will visually block the ad but not the ad tracking.

martijnvds | 14 hours ago

Exactly how the big players like it.

TheDong | 17 hours ago

The number of drive-by malware attacks I've gotten from having uBO installed is zero.

The number of drive-by malware attacks I've gotten from ad-slots on various websites is non-zero, and the number of accidental "click on ad, end up on malicious website" incidents is even higher.

For me, the math on uBO has been working out great so far.

mahboi | 17 hours ago

There's almost no overlap between sites you shouldn't mess with and sites that need ad-blocking. Like Chase isn't running third-party ads, and Google doesn't need login.

TheDong | 17 hours ago

Quite a few of my "accidentally click on ad, end up on malicious site" are from perfectly trustworthy sites.

For example, google search result ads. Where an attacker pays more money than the real company. Wildly common.

And even if you don't login to google, you still get served potentially malicious links to accidentally click on. Stripping those out with an ad-blocker is better than hoping you never misclick.

mahboi | 7 hours ago

Google was my example of a site you do want ad-blocking on and also don't mind extensions having access to

fc417fc802 | 14 hours ago

You should double check what your bank is doing under the hood. I can't speak to chase but mine would load lots of questionable third party garbage if it weren't for umatrix.

cdman | 16 hours ago

Malicious ads on Google.com (as in: on the top of the search results) are routine occurrence. Blocking ads for you and your loved ones is the second best thing you can do to keep them safe (after keeping your OS/software updated).

actualwitch | 20 hours ago

> a similar but less problematic issue also affected the MV3 version, uBO Lite

madeofpalk | 19 hours ago

So, Manifest V3 wins again

fc417fc802 | 20 hours ago

Manifest v3 is the adblock and sanitization equivalent of encasing your computer in cement and then declaring it secure.

NooneAtAll5 | 20 hours ago

> encasing your computer in cement

mafia execution?