This should not be considered an LLM issue. It's a user experience issue.
Mozilla has made changes that happened by default before. Often I have had to find the setting to put it back to how I wanted. I remember when it moved the URL bar to the bottom.
I don't think it is always an easy call to make. Tabs were a significant user experience improvement, but hiding it behind an opt-in would have limited it to people who knew about it.
I use Firefox as main on desktop and mobile. I have noticed messages on upgrade pointing to LLM features. I haven't engaged with them an from thereon haven't noticed any change because of them.
Saying there are reports of excessive memory or CPU use isn't terribly useful without references to those reports. One such report posted on HN was shown to have been unrelated to the LLM.
Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
LLMs are garbage and they add nothing to the browsing experience.
> Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
You can control this option with a setting. It seems like it would be really really easy to just test this. As a result I can't see any reason to doubt this by default or apply a legalistic evidentiary standard when considering it.
> LLMs are garbage and they add nothing to the browsing experience.
The builtin translation feature [0] is LLM-based [1], and that adds a ton to my browsing experience, since it's made web pages in other languages accessible to me.
[1]: According to Wikipedia, "A large language model (LLM) is a language model trained with self-supervised machine learning on a vast amount of text, designed for natural language processing tasks" [2]. The translation code is transformer/RNN-based and trained on raw texts [3, 4], and translation definitely qualifies as a natural language processing task, meaning that the translation feature is LLM-based.
What evidence do you have that it wasn't tested? Or are you just making an assumption based on one article? Did you do any follow up searching? It might have led you to this:
NMT doesn't "contain" tranformers and deep RNNs, it can use them. LLMs use a transformer architecture, not everything using a transformer architecture is an LLM. NMT can actually use an LLM, but that's not the case according to the documentation you linked, they use a parallel dataset to train their models.
> Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
AFIK no
nearly all AI features only do any compute if you explicitly ask them to do stuff
the only exception is that when you create a tab group (and only then!) it will auto recommend a name. And while I don't find a very useful for me (and I use tab groups a ton) I also have observed that non technical users get confused with unnamed tab groups and the UI around them so I really understand why they do that. Also while it's "technically" a LLM it's tiny and fast to run and could also have been implemented with just embeddings at a slightly loss of quality (and again only runs if you create a tab group and not at any other time and not if you computer has less then 3GiB of RAM etc. etc.(yes FF disables AI features on very low RAM devices AFIK))
Put people hear AI think LLM get angry and stop thinking.
Or considering that maybe a non negligible amount of people outside of their bubble want exactly that and might leave your browser if you don't supply that.
I understand why Mozilla have started implementing these features, it seems to have more mass market appeal than not - look at how popular “AI” powered browsers have become.
But boy does it not add extra effort removing these features every time there’s a new roll-out and it’s not done the best way IMO.
I feel as if these features would go down better if Mozilla actually notified the user that they’re available and then offered whether to enable them or not (could have them enabled by default for new users). That way you’re still giving a choice, but in a more respectful manner.
If anyone is interested I’ve gutted all the more obscene stuff out of Waterfox and have instead left the useful ones such a ML translation, which is opt-in.
Related: I feel like onboarding is a lost art, more software should bring back software wizards and UI tours. Feels like you somehow have to intuitively know how something works (unlikely) or do a web search on how to use everything instead of having it shown to you nicely.
> Related: I feel like onboarding is a lost art, more software should bring back software wizards and UI tours.
Yes, please! We use Google's online office programs at work and every time it has so far popped up a notification about a new feature I immediately dismissed it by the act of actually doing the work I opened the tab to do. Then I have no idea how to find out what that feature was again, as the popup notification was dismissed.
I'm torn because I probably do want to know what new, relevant features are, but I opened the app for a reason, so I want the tour to get out of my way so I can do my work.
So true about onboarding. The trend towards trying to make all interfaces simple and immediately accessible has led to lots of functionality being deeply hidden. Of course you don't want the interface to be insanely complicated but you also don't want to make it so simple it's limiting.
Sort of related, after reading this I went and checked the Waterfox reddit and saw some people complaining about recent changes. I agree with a person there that one of the most important things is not changing. One of the reasons I use Waterfox is to not be subject to the caprices of Mozilla. I just want the same interface I've been using since back in the days of like Firefox 4.0. If there are changes, they can be introduced in an opt-in, reversible way as you suggest. But the default assumption should be "don't break users' workflows by changing behavior".
I appreciate all you do with Waterfox! I've been using it for years now.
They're all easily disabled in the GUI itself. The article is exaggerating, the closest argument is that it enables itself by default when it first updated which is fair, but they're easy to disable within the menu itself.
At least the local AI features, being able to translate or summarize a page without sending it off to Google, seem like they'd appeal to this crowd the most.
I don't really understand the article's complaints (like "high cpu & ram usage with firefox local ai features" and "forcing [...] without asking the user") when to my understanding they're something you have to intentionally decide to use and are not using any resources otherwise - unless there's some feature I've missed.
Any AI features, helpful or no, seems to me like something that should exist in the form of an extension, not baked into the browser. Or, if baked in, should be opt in, not opt out by default, as regardless of where you fall on it, AI is a divisive topic and people have strong opinions on it. Especially the people in Firefoxes remaining small audience that tend to be more technically literate and privacy/security conscious.
I have no problem with it announcing new features and asking if I wish to enable them after an update, but I'd really rather prefer it not to force enable new features (regardless of proximity/relevance to AI) whether it tells me about them up front - and especially don't want it to if they aren't even notifying me of it.
Local translation, sure. Summarization? No... For one thing, I don't use anything for that (does Google have a "Summarize This" feature now?) But unless I missed it, I don't even think this article disables the local translation feature. I'm pretty sure that is controlled by `browser.translations.enable`.
> At least the local AI features, being able to translate or summarize a page without sending it off to Google, seem like they'd appeal to this crowd the most.
Yes I would like more local translation support! (Faster, better, more languages, sometimes it fails on mobile for unknown reasons). Also Firefox history could use refresh (like suggested here https://community.brave.app/t/improvements-for-browser-histo... and 5 can be AI enhanced)
That being said, I'm a FF user interested in exploring what LLM features in webbrowsers could bring. I would hate it if lack of them would make me switch to chromium-based browsers. So I'm happy Mozilla is exploring these.
It annoys me that I have to disable this alongside the telemetry I already had to disable (and their Home page crap).
Telemetry should always be opt in not opt out - I don't care how you justify it but especially I don't care when you've marketed yourself with "Firefox is built with privacy and protection as the default."[1]
When it comes to on by default - If what you are doing amounts to "I Am Altering the Deal, Pray I Don't Alter It Any Further." then you might have to ask yourself whether you should be doing it.
I switched to Vivaldi because I was fed up with Firefox.
I'm not thrilled it isn't open source. But it works, it has the best out of the box experience I've seen in any browser, and they don't shove AI features in my face.
Appealing to that crowd is how they grew their market share the first time around. The techies came first, trust was established, and then they recommended it to everyone else.
Do you really think AI features are going to move the needle here? Is your grandma going to think "Wow, Firefox can summarize my emails!" and switch web browsers?
Most of Firefox user base has always been on Windows, not Linux. What OS do you think the "techies" that promoted Firefox to replace IE in the first place were running?
Sure maybe 20 years ago. But back then Linux's userbase was also on Windows, because desktop Linux hadn't really become usable yet. I think nowadays Firefox's marketshare is a lot higher on e.g. Ubuntu (where it's the default) than it is on Windows (where Edge is the default).
Optimal marketing strategies change over time. What may have worked 20 years, may not be the right approach today. Many businesses struggle with this. They remember what they did to initial gain success, but that marketing route is no longer effective so they fail to figure out how to reach new people and eventually die.
>Do you really think AI features are going to move the needle here?
Most prominent thing is chat in sidebar. It's iframe + a few shortcuts. Optional, harmless, using zero resources. Actually quite convenient.
Another is perplexity being one of the search providers. This is literally config not code. I wonder how many people actually removed or even looked at the list of the default search providers before that.
I think only real one is ml naming for tabs. Just meh.
Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
Way to move the goalposts. The argument is not about the usefulness, it's about baked-in on by default features for what are obvious extension workflows. About the only marketing advantage FF has at this point is customizable and lack of dirty tricks. Mozilla seems confusingly desperate to sell out here asap
They would need to finance a lot less development if they stopped changing things and just worked on maintaining a limited core functionality including a nice extension system. That's it. Most of their "development" over the last 10 years has been a total waste of time.
I don't know, do you have evidence to the contrary?
Either way, all I'm saying is when I look at the UI changes that have been made to Firefox over a period of years, I believe their value to me is closer to zero than their development cost is.
Agreed; I wish they would expose how to set your own chat provider more easily than about:config, but the sidebar chat is easily my most used unique "UX" Firefox feature.
I'd like to see (opt-in) automatic grouping. The kinda-sorta grouping it does still requires you to manually engage with it.
Personally I don't care - it could be first-party extension.
But you could bring up the same argument about vertical tabs, tab groups, Firefox Sync or many other things that interfere with the browser on much deeper level than shortcut to open page in the sidebar.
Is there argument to single out this specific feature apart from "AI bad"?
> Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
What made you jump to that conclusion? My guess about someone who's using a non-mainstream browser and figuring out how to configure it to their liking is that they're likely also using AI in more ways than the standard Chat-Webinterface, eg Agents, CLI tools, MCP,... To give an analogy, rolling your eyes over brainrot memes isn't denying the usefulness of smartphones or messengers either. The underlying sentiment is being critical of things that get pushed down our throats through A/B-optimized patterns that ultimately serve other interests than your own, profits or darker.
> Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
This is a remarkably useless argument; "any" could be anything whatsoever, for example that the fuzzy logic of some prior AI craze ended up in certain rice cookers, while ignoring that the remaining 99.8881118881118883479075520881451666355133056640625 (or so) percent of AI is some combination of grift, wishful thinking, or both.
Mozilla seems to be an organization that is metrics driven to a degree that would make Robert MacNamarra blush.
I think this is why they keep shoving new features at users whether they want them or not, making them incredibly difficult to disable, rather than presenting an option try something new, or even making opting out of features easy and intuitive.
The latter two would lead to fewer users of the feature, which means it risks being removed for not being used by most users. Not to mention having an easy opt-out functionality means its usage can be tracked, which could generate unwanted statistics and make a stakeholder lose face.
There's a checkbox on whether you want to use it or not in the settings page, does this not change these settings?
I don't feel opposed to them changing the browser in principle--certainly there have been many improvements to web browsers over the years. Is privacy the concern here?
If the checkbox you're referring to is the "Use AI to suggest tabs and a name for tab groups" one, then I can't see what setting it changes. It's not the browser.ml.enable flag. I tried unchecking it, restarting the browser, and that flag was unaffected. This is in version 144.0.2.
Searching for "AI" shows one other setting: "Quickly access bookmarks, tabs from your phone, AI chatbots, and more without leaving your main view." But I'd already disabled that apparently. Despite that, there are plenty of flags that were enabled mentioned in the article.
Last I checked there wasn’t and you still had to fiddle with a few about:config options to actually turn off all the ai stuff. I would be fine with it if it was just a settings page rather than hidden settings.
I know I'm a broken record on this topic, but messing with prefs directly is not a supported way of configuring Firefox, so if you do this, you shouldn't complain that Mozilla changed the prefs and reenabled these features.
It's of course totally fine to be annoyed if Firefox doesn't have a supported preference to disable them (though from the discussion below it sounds like they do) or if you set that preference and then they change it.
> > you shouldn't complain that Mozilla changed the prefs and reenabled these features.
> Then they should provide a reliable way of getting what I want.
As I said in the comment you are responding to, it's perfectly legitimate to complain if Mozilla didn't provide a setting to disable these features (though it seems they have done so in at least some cases). However, if they don't do so, and you decide to disable them in some unsupported way, you shouldn't be surprised that you get unexpected results.
> This notion that software developers know better than me is poisonous.
I don't think this is really a reasonable proposition as a general matter. A browser is a very complicated piece of software and each individual configuration point is additional development and testing burden, both separately, and in combination with other configuration points. Developers need to make all kinds of decisions in order to manage engineering complexity. In some cases, where developers know that people will want to configure things, they'll provide configuration options but it's not practical to do so for every feature, and inevitably in some cases they'll get it wrong (see "legitimate to complain" above).
From the posted article: "The main problem with this is users are having this forced on them with no gui option to disable these features."
> and you decide to disable them in some unsupported way
The preferences file is "unspported?" Why? Is firefox intentionally making their browser hard to distribute? All the settings, even in the UI, have simple textual names. Is there some reason they need to have this support envelope be so underwhelming? What advantage does that actually bring? If there is none, then why _shouldn't_ I complain? The alternative is I can stop using the project and recommend to everyone that they do so as well.
Here's the awesome support you get through the UI:
"If you are also dealing with CPU spikes and battery drain from Firefox's new AI features, you can disable them through the browser's advanced settings. Head to about:config in a new tab, accept the risk warning, and use the search bar to find the controls. To kill the AI chatbot feature, search for browser.ml.chat.enabled and set it to false. To stop smart tab grouping, search for browser.tabs.groups.smart.enabled and set it to false."
Give me a break.
> I don't think this is really a reasonable proposition as a general matter.
Then I won't use your software. I will recommend against it. I will ensure that everyone understands it's faults. That's the only reasonable response as an irritated user.
A supported way would imply guarantees of compatibility and stability.
These sort of perfs in user.js or about:config are clearly labeled as not having those guarantees and therefore "unsupported".
As someone who extensively customizes the look, feel and way that firefox works I think this makes sense. I of course would like mozilla to make a lot more options "officially" supported, especially turning off anything that contacts a remote server without explicitly wanted.
It's boggling my mind that FireFox is 2% of the browser market now. I'm hoping that includes mobile where it makes a lot more sense. But even so, really sad. It is actually a great browser (putting aside this issue of AI features being pushed) and I actually do have much greater trust that it is respecting my privacy over Chrome and Edge.
I'm not sure if I ever used any LLM features from Firefox besides exploring, but so far it seems great they're trying local approach instead of sending it to some "partner".
In the comment section here I see lot of people complaining about the fact it's enabled by default as well as some concerns about resource usage. Could someone experienced in desktop app architecture explain if disabling them functionalities makes Firefox that much faster or using less resouces? I'd assume that those functionalities are kind of loaded on demand?
"AI" became such a keyword that seem to instantly give either positive or negative response, it's also an advertised feature of every second app with many of them just forcing AI into you just because of hype. This doesn't seem to be a case in Firefox - so I highly disagree with the title - the features are there but they don't go into your way if you don't want them, therefore it's easy to just use it, only when needed
I also think that we in the long run will probably let machines do most tedious browsing for us-- digesting ad-ridden websites, digesting interfaces. The LLM navigates the actual web, presented to maximize revenue and maximize user engagement, time spent on the website etc., but we only see actual content, carefully arranged to be as comprehensible as possible, and if we want to communicate with somebody through a website controlled by others we formulate the message and the LLM submits it.
I'm so confused, they have a whole page that starts with how they run everything on device[0]. But the page ends telling you how great it is that you can choose your AI service which then leads to a page where ... all the options are hosted AI providers and half the page is about you have to read their terms of service and they all invade your privacy[1]. I stupidly enabled Gemini at some point and now there's no obvious way in the UI to disable it. I can only switch to a different hosted service provider.
I searched in settings for AI, model, LLM, nada.
Whatever good things they did here they totally screwed up the comms and UI execution of it.
I think that (optional) AI features are fine, but the model selection only shows the big providers by default. You can enable local LLM support under about:config -> browser.ml.chat.hideLocalhost (set to false). It will check localhost:8080, which is not a great choice for a fixed port. I'd prefer if I could just point it at any OpenAI-compatible endpoint. Either way, seems to work well when serving a local model with llama.cpp.
I've been using Vivaldi alongside Firefox anyways for the edge cases where non-chromium browsers refused to work but I've switched fully now because of the latest round of AI crap. At least Vivaldi has explicitly committed themselves to keep it out of the browser.
Sharing my configs, hopefully they'll be helpful to someone. I don't even use Firefox but I keep the settings updated (prefs.js can be modified by Mozilla/Firefox, user.js does not change).
NOTE: Some settings might block too much, edit and use as you please.
I've found the "Copy text from image" menu command to be useful from time to time, and the features I'm not interested in seemed easy to turn off/dismiss/ignore?
They’re opt-out and can be disabled in the settings or fully disabled in about:config. Definitely annoying but not enough to make me switch to a Chromium based browser.
> They’re opt-out and can be disabled in the settings or fully disabled in about:config.
Any setting where there's a difference between "disabled" and "fully disabled" is user hostile. And, for a company that advertises itself as all about respecting the user, Mozilla sure does love their user-hostile decisions.
The built in LLM translation is great. I'm leaving that enabled. It's useful and it's a perfect application for local LLM to be better than the equivalent corporate services.
I disable a ton in default FF and even run the unbranded versions so that it's not trialware (FF branded builds all expire when their baked in add-ons CA TLS certs expire). But the LLM translation? That's finally a good feature.
It isn't clear what browser.ml.chat and browser.ml.pageAssist are associated with in terms of features. Does anyone know? I tried disabling all shown in the write-up and local LLM translation still seems to work so I assume it's something else.
Honestly, I haven't even noticed LLM features in the desktop version. But I found it really annoying in the mobile (iOS) app. There's a "Summarize Page" feature that's adjacent to "Find in Page." It's easy to mis-tap when you're just trying to search for a term on the page. It's also activated by the shake gesture, which can happen accidentally.
Three dots -> Settings -> Page Summaries to disable that.
It's important to remember that if people think of Firefox as a way to escape LLMs being pushed on them in Chrome, they might accidentally attract users from Chrome. It's the only possible reason why this wouldn't be a plugin. Why else would a user-focused browser consciously come up with strategies that attack user preferences?
You just don't understand the eleven-dimensional chess that you have to play to get from 30% marketshare to 2% marketshare. They have to think they have a winning strategy, judging by the way they talk to everybody who criticizes their decisions.
sure the smart tab group features are kind useless (but tab groups are grate)
but I don't understand why people make such a deal about it
it looks a loot like "I don't like this so it mustn't even exist even if some people like it" mentality
like
- all of this features are opt in, they don't just randomly send your data somewhere or randomly eat your CPI resources (without you having enabled them)
- the smart tab group features(1) are the only very visible and kind dump/useless one, and you can disable them without tweaking internal configs unser "Addons > On-device AI" (but it won't disable the button in the group tabs menu which is used to enable that feature in the first place)
- for many users side translations is an important must have feature, it's one of the AI features, and having a local on-device no data send to servers translation is very desirable (to be fair US users probably don't care nor realize it matters, but it really matters) (To be fair the OP didn't disable this either)
- for increasingly many users integration with chat bots (e.g. ChatGPT) is a must have browser feature(2). The integration is also only visible in a context menu(3) and hidden in the a selector for the side bar and needs to be explicitly setup before it does anything. So a feature some people want and treat as must have and now can have and for everyone else it's pretty much out of the way and not pushed onto you at all. So why insists that it's not okay to have it even in a selector hardly any power users will ever use (because the other options there are bookmarks, and history, both with well known widely used keyboard shortcuts)
- the preview was a "Labs" (i.e. experimental preview) feature you have to explicitly enable and then use the right hot key to use, and it's one of this "some people expect this feature in 2025" features (but not sure what happened to it, it's neither in Labs nor enabled in any FF browser I have, so maybe they discontinued it)
(1): They IMHO are pretty dump and seem more like a way to setup the whole infrastructure around them with a trivial to implement feature then a in depth feature mozilla cares about. I guess.
(2): On HN it often seems like only enthusiasts and not many non-tech people use ChatGPT and co. _That is not true at all_. When it comes to "simple" ChatGPT usage far more non technical users use it then technical one. Like for dump stuff like shopping list, holiday check lists, etc. The problem OpenAI has isn't that they don't have a lot of users, it's that they need a absurd amount of high paying users. But the tasks most non-tech people use generic ChatGPT for aren't worth a lot. Like even the Plus plan is a "only if you are wealthy enough to burn 20 bucks every month without caring" option in such a context and the Pro plan just absurd.
(3): Is anyone except non technical users still using browsers context menus in a way where a "Ask AI Chat-Bot" entry matters???? Like copy, past, undo, redo, select all have trivial short cuts. Language/spell check is rarely touched and inspect has a short cut where you press Q after right clicking.
deleated from most recent linux install on a new to me desktop.
vile
I just want to work(create stuff), there is nothing that any "service" can do for me.
the relentles attempts to get in the way of sending and recieving information is mindbogling, and I am going to be getting an analog telephone adapter, so that I can start useing fax, over cellular data, and making paper plans ,taking pictures, and converting to pdf's to avoid the wasted time and cost of software solutions
the surreal part is that the LLM's have found me, and are recomending me to people looking for custom metal, so much for a low profile there, and somehow my number got out on indeed, and I got a wave of people calling, applying for "the job" but bitterly complaining that the app they downloaded as part of that, wasn't working right.
fine, ok, whatever.
This should not be considered an LLM issue. It's a user experience issue.
Mozilla has made changes that happened by default before. Often I have had to find the setting to put it back to how I wanted. I remember when it moved the URL bar to the bottom.
I don't think it is always an easy call to make. Tabs were a significant user experience improvement, but hiding it behind an opt-in would have limited it to people who knew about it.
I use Firefox as main on desktop and mobile. I have noticed messages on upgrade pointing to LLM features. I haven't engaged with them an from thereon haven't noticed any change because of them.
Saying there are reports of excessive memory or CPU use isn't terribly useful without references to those reports. One such report posted on HN was shown to have been unrelated to the LLM.
Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
> This should not be considered an LLM issue.
LLMs are garbage and they add nothing to the browsing experience.
> Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
You can control this option with a setting. It seems like it would be really really easy to just test this. As a result I can't see any reason to doubt this by default or apply a legalistic evidentiary standard when considering it.
> LLMs are garbage and they add nothing to the browsing experience.
The builtin translation feature [0] is LLM-based [1], and that adds a ton to my browsing experience, since it's made web pages in other languages accessible to me.
[0]: https://support.mozilla.org/en-US/kb/website-translation
[1]: According to Wikipedia, "A large language model (LLM) is a language model trained with self-supervised machine learning on a vast amount of text, designed for natural language processing tasks" [2]. The translation code is transformer/RNN-based and trained on raw texts [3, 4], and translation definitely qualifies as a natural language processing task, meaning that the translation feature is LLM-based.
[2]: https://en.wikipedia.org/wiki/Large_language_model
[3]: https://www.youtube.com/watch?v=J06koBcfm5w
[4]: https://marian-nmt.github.io/features/
It's easy to test, therefore there is no need to test it? I don't see how that follows. If you haven't tested it then you haven't tested it.
> therefore there is no need to test it?
What evidence do you have that it wasn't tested? Or are you just making an assumption based on one article? Did you do any follow up searching? It might have led you to this:
https://www.neowin.net/news/mozilla-under-fire-for-firefox-a...
Which says.
Update (August 13, 2025, 03:57 GMT): While the community correctly identified a performance issue, their attribution of the cause was mistaken.
.
And is also not using an LLM. It's neural machine translation.
NMT is a category containing both transformers and deep RNN. The Mozilla translation models are transformer LLM NMTs trained via Marian https://marian-nmt.github.io/ (ref: https://github.com/mozilla/translations/blob/main/docs/READM...)
NMT doesn't "contain" tranformers and deep RNNs, it can use them. LLMs use a transformer architecture, not everything using a transformer architecture is an LLM. NMT can actually use an LLM, but that's not the case according to the documentation you linked, they use a parallel dataset to train their models.
> they use a parallel dataset to train their models
If you want to be pedantic you should look up the LLM definition.
Care to explain why?
> Are there any reports actually showing degradation because of LLMs rather than post hoc ergo propter hoc?
AFIK no
nearly all AI features only do any compute if you explicitly ask them to do stuff
the only exception is that when you create a tab group (and only then!) it will auto recommend a name. And while I don't find a very useful for me (and I use tab groups a ton) I also have observed that non technical users get confused with unnamed tab groups and the UI around them so I really understand why they do that. Also while it's "technically" a LLM it's tiny and fast to run and could also have been implemented with just embeddings at a slightly loss of quality (and again only runs if you create a tab group and not at any other time and not if you computer has less then 3GiB of RAM etc. etc.(yes FF disables AI features on very low RAM devices AFIK))
Put people hear AI think LLM get angry and stop thinking.
Or considering that maybe a non negligible amount of people outside of their bubble want exactly that and might leave your browser if you don't supply that.
I understand why Mozilla have started implementing these features, it seems to have more mass market appeal than not - look at how popular “AI” powered browsers have become.
But boy does it not add extra effort removing these features every time there’s a new roll-out and it’s not done the best way IMO. I feel as if these features would go down better if Mozilla actually notified the user that they’re available and then offered whether to enable them or not (could have them enabled by default for new users). That way you’re still giving a choice, but in a more respectful manner.
If anyone is interested I’ve gutted all the more obscene stuff out of Waterfox and have instead left the useful ones such a ML translation, which is opt-in.
Related: I feel like onboarding is a lost art, more software should bring back software wizards and UI tours. Feels like you somehow have to intuitively know how something works (unlikely) or do a web search on how to use everything instead of having it shown to you nicely.
> Related: I feel like onboarding is a lost art, more software should bring back software wizards and UI tours.
Yes, please! We use Google's online office programs at work and every time it has so far popped up a notification about a new feature I immediately dismissed it by the act of actually doing the work I opened the tab to do. Then I have no idea how to find out what that feature was again, as the popup notification was dismissed.
I'm torn because I probably do want to know what new, relevant features are, but I opened the app for a reason, so I want the tour to get out of my way so I can do my work.
All I'd really want is a button for such a tour up in the corner by the "Ask Gemini" button. Out of the way, but still there when I've got the time.
Seriously. I'm the type of person to read the what's new, but they're temporally placed for minimal engagement with them
So true about onboarding. The trend towards trying to make all interfaces simple and immediately accessible has led to lots of functionality being deeply hidden. Of course you don't want the interface to be insanely complicated but you also don't want to make it so simple it's limiting.
Sort of related, after reading this I went and checked the Waterfox reddit and saw some people complaining about recent changes. I agree with a person there that one of the most important things is not changing. One of the reasons I use Waterfox is to not be subject to the caprices of Mozilla. I just want the same interface I've been using since back in the days of like Firefox 4.0. If there are changes, they can be introduced in an opt-in, reversible way as you suggest. But the default assumption should be "don't break users' workflows by changing behavior".
I appreciate all you do with Waterfox! I've been using it for years now.
> look at how popular “AI” powered browsers have become.
Is there an active user market for browsers? If there isn't then this analysis is useless.
There is an active market for users and 'AI' is the new big thing to feel them in.
> ...look at how popular “AI” powered browsers have become.
What? I have yet to meet a single person who has any interest in "AI powered browsers"
They're all easily disabled in the GUI itself. The article is exaggerating, the closest argument is that it enables itself by default when it first updated which is fair, but they're easy to disable within the menu itself.
Anything enabled by default without prompting in an update is usually against the user.
Where? I just looked in Settings.
Right click on the icon → remove.
The same goes for right clicking on the page → ask AI → remove.
How? about:config doesn’t count as “the GUI”, imo.
Mozilla sure is going out of their way to alienate their remaining users. This is going to be Pocket 2.0.
Who asked for this? Who wants it? Certainly not the Linux / open-source crowd, and they're just about the only ones who are keeping Firefox alive.
If there's anybody from Mozilla or the Firefox dev team in this thread, I'd be interested to hear the thinking behind this addition.
> Certainly not the Linux / open-source crowd
At least the local AI features, being able to translate or summarize a page without sending it off to Google, seem like they'd appeal to this crowd the most.
I don't really understand the article's complaints (like "high cpu & ram usage with firefox local ai features" and "forcing [...] without asking the user") when to my understanding they're something you have to intentionally decide to use and are not using any resources otherwise - unless there's some feature I've missed.
Any AI features, helpful or no, seems to me like something that should exist in the form of an extension, not baked into the browser. Or, if baked in, should be opt in, not opt out by default, as regardless of where you fall on it, AI is a divisive topic and people have strong opinions on it. Especially the people in Firefoxes remaining small audience that tend to be more technically literate and privacy/security conscious.
I have no problem with it announcing new features and asking if I wish to enable them after an update, but I'd really rather prefer it not to force enable new features (regardless of proximity/relevance to AI) whether it tells me about them up front - and especially don't want it to if they aren't even notifying me of it.
Local translation, sure. Summarization? No... For one thing, I don't use anything for that (does Google have a "Summarize This" feature now?) But unless I missed it, I don't even think this article disables the local translation feature. I'm pretty sure that is controlled by `browser.translations.enable`.
> At least the local AI features, being able to translate or summarize a page without sending it off to Google, seem like they'd appeal to this crowd the most.
Yes I would like more local translation support! (Faster, better, more languages, sometimes it fails on mobile for unknown reasons). Also Firefox history could use refresh (like suggested here https://community.brave.app/t/improvements-for-browser-histo... and 5 can be AI enhanced)
It should be opt-in to avoid alienating users.
That being said, I'm a FF user interested in exploring what LLM features in webbrowsers could bring. I would hate it if lack of them would make me switch to chromium-based browsers. So I'm happy Mozilla is exploring these.
It annoys me that I have to disable this alongside the telemetry I already had to disable (and their Home page crap).
Telemetry should always be opt in not opt out - I don't care how you justify it but especially I don't care when you've marketed yourself with "Firefox is built with privacy and protection as the default."[1]
[1] https://www.firefox.com/en-US/user-privacy/
When it comes to on by default - If what you are doing amounts to "I Am Altering the Deal, Pray I Don't Alter It Any Further." then you might have to ask yourself whether you should be doing it.
I switched to Vivaldi because I was fed up with Firefox.
I'm not thrilled it isn't open source. But it works, it has the best out of the box experience I've seen in any browser, and they don't shove AI features in my face.
Yeah. As a Mozilla/FF user for as long as I have memory ... unwanted "features" are about the only way to get me stop using this browser.
>Certainly not the Linux / open-source crowd
Appealing to that crowd is not big enough to grow it's market share from 2%.
Appealing to that crowd is how they grew their market share the first time around. The techies came first, trust was established, and then they recommended it to everyone else.
Do you really think AI features are going to move the needle here? Is your grandma going to think "Wow, Firefox can summarize my emails!" and switch web browsers?
Most of Firefox user base has always been on Windows, not Linux. What OS do you think the "techies" that promoted Firefox to replace IE in the first place were running?
Sure maybe 20 years ago. But back then Linux's userbase was also on Windows, because desktop Linux hadn't really become usable yet. I think nowadays Firefox's marketshare is a lot higher on e.g. Ubuntu (where it's the default) than it is on Windows (where Edge is the default).
According to Mozilla's own data at https://data.firefox.com/dashboard/hardware Windows (7, 10 & 11) make up 84% of their user base.
Optimal marketing strategies change over time. What may have worked 20 years, may not be the right approach today. Many businesses struggle with this. They remember what they did to initial gain success, but that marketing route is no longer effective so they fail to figure out how to reach new people and eventually die.
>Do you really think AI features are going to move the needle here?
It's better than doing nothing at least.
I want this. I'm Firefox user for years btw.
Most prominent thing is chat in sidebar. It's iframe + a few shortcuts. Optional, harmless, using zero resources. Actually quite convenient.
Another is perplexity being one of the search providers. This is literally config not code. I wonder how many people actually removed or even looked at the list of the default search providers before that.
I think only real one is ml naming for tabs. Just meh.
Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
Way to move the goalposts. The argument is not about the usefulness, it's about baked-in on by default features for what are obvious extension workflows. About the only marketing advantage FF has at this point is customizable and lack of dirty tricks. Mozilla seems confusingly desperate to sell out here asap
Why this is obvious extension workflow and vertical tabs or FF Sync aren't?
How do you propose they finance development if we consider simple integrations with zero privacy or customization impact "selling out"?
Those are also obvious extension workflow.
They would need to finance a lot less development if they stopped changing things and just worked on maintaining a limited core functionality including a nice extension system. That's it. Most of their "development" over the last 10 years has been a total waste of time.
You really think these things are requiring significant development budget, compared to keeping up with things like wasm, webgpu, grid/flex layouts?
Can you really say with a straight face, that not doing vertical tabs or chat shortcut would allow them to operate on much smaller budget?
I don't know, do you have evidence to the contrary?
Either way, all I'm saying is when I look at the UI changes that have been made to Firefox over a period of years, I believe their value to me is closer to zero than their development cost is.
Agreed; I wish they would expose how to set your own chat provider more easily than about:config, but the sidebar chat is easily my most used unique "UX" Firefox feature.
I'd like to see (opt-in) automatic grouping. The kinda-sorta grouping it does still requires you to manually engage with it.
Any reason why extensions couldn’t fill these niches?
Personally I don't care - it could be first-party extension.
But you could bring up the same argument about vertical tabs, tab groups, Firefox Sync or many other things that interfere with the browser on much deeper level than shortcut to open page in the sidebar.
Is there argument to single out this specific feature apart from "AI bad"?
> Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
What made you jump to that conclusion? My guess about someone who's using a non-mainstream browser and figuring out how to configure it to their liking is that they're likely also using AI in more ways than the standard Chat-Webinterface, eg Agents, CLI tools, MCP,... To give an analogy, rolling your eyes over brainrot memes isn't denying the usefulness of smartphones or messengers either. The underlying sentiment is being critical of things that get pushed down our throats through A/B-optimized patterns that ultimately serve other interests than your own, profits or darker.
Making a very simple feature and enabling it, with option to disable it is no pushing down the throat, nor dark pattern.
Please don't compare it to Google or Microsoft products that give you "Enable now/Remind later" or just no option at all.
It's the beginning, A/B-testing takes time...
> Honestly people who deny any usefulness of AI are getting dangerously close to flat-earthers by now.
This is a remarkably useless argument; "any" could be anything whatsoever, for example that the fuzzy logic of some prior AI craze ended up in certain rice cookers, while ignoring that the remaining 99.8881118881118883479075520881451666355133056640625 (or so) percent of AI is some combination of grift, wishful thinking, or both.
Mozilla seems to be an organization that is metrics driven to a degree that would make Robert MacNamarra blush.
I think this is why they keep shoving new features at users whether they want them or not, making them incredibly difficult to disable, rather than presenting an option try something new, or even making opting out of features easy and intuitive.
The latter two would lead to fewer users of the feature, which means it risks being removed for not being used by most users. Not to mention having an easy opt-out functionality means its usage can be tracked, which could generate unwanted statistics and make a stakeholder lose face.
There's a checkbox on whether you want to use it or not in the settings page, does this not change these settings?
I don't feel opposed to them changing the browser in principle--certainly there have been many improvements to web browsers over the years. Is privacy the concern here?
If the checkbox you're referring to is the "Use AI to suggest tabs and a name for tab groups" one, then I can't see what setting it changes. It's not the browser.ml.enable flag. I tried unchecking it, restarting the browser, and that flag was unaffected. This is in version 144.0.2.
Searching for "AI" shows one other setting: "Quickly access bookmarks, tabs from your phone, AI chatbots, and more without leaving your main view." But I'd already disabled that apparently. Despite that, there are plenty of flags that were enabled mentioned in the article.
Last I checked there wasn’t and you still had to fiddle with a few about:config options to actually turn off all the ai stuff. I would be fine with it if it was just a settings page rather than hidden settings.
For people using NixOS, you can wrap these configs in extraPrefs in wrapFirefox firefox-unwrapped:
Example: Put this expression(using lockPref to hardcode the config values) in environment.systemPackages(assuming "with pkgs"):
I know I'm a broken record on this topic, but messing with prefs directly is not a supported way of configuring Firefox, so if you do this, you shouldn't complain that Mozilla changed the prefs and reenabled these features.
It's of course totally fine to be annoyed if Firefox doesn't have a supported preference to disable them (though from the discussion below it sounds like they do) or if you set that preference and then they change it.
> is not a supported way of configuring Firefox
What is the supported way?
> you shouldn't complain that Mozilla changed the prefs and reenabled these features.
Then they should provide a reliable way of getting what I want. This notion that software developers know better than me is poisonous.
> What is the supported way?
The settings UI.
> > you shouldn't complain that Mozilla changed the prefs and reenabled these features.
> Then they should provide a reliable way of getting what I want.
As I said in the comment you are responding to, it's perfectly legitimate to complain if Mozilla didn't provide a setting to disable these features (though it seems they have done so in at least some cases). However, if they don't do so, and you decide to disable them in some unsupported way, you shouldn't be surprised that you get unexpected results.
> This notion that software developers know better than me is poisonous.
I don't think this is really a reasonable proposition as a general matter. A browser is a very complicated piece of software and each individual configuration point is additional development and testing burden, both separately, and in combination with other configuration points. Developers need to make all kinds of decisions in order to manage engineering complexity. In some cases, where developers know that people will want to configure things, they'll provide configuration options but it's not practical to do so for every feature, and inevitably in some cases they'll get it wrong (see "legitimate to complain" above).
> The settings UI.
From the posted article: "The main problem with this is users are having this forced on them with no gui option to disable these features."
> and you decide to disable them in some unsupported way
The preferences file is "unspported?" Why? Is firefox intentionally making their browser hard to distribute? All the settings, even in the UI, have simple textual names. Is there some reason they need to have this support envelope be so underwhelming? What advantage does that actually bring? If there is none, then why _shouldn't_ I complain? The alternative is I can stop using the project and recommend to everyone that they do so as well.
Here's the awesome support you get through the UI:
"If you are also dealing with CPU spikes and battery drain from Firefox's new AI features, you can disable them through the browser's advanced settings. Head to about:config in a new tab, accept the risk warning, and use the search bar to find the controls. To kill the AI chatbot feature, search for browser.ml.chat.enabled and set it to false. To stop smart tab grouping, search for browser.tabs.groups.smart.enabled and set it to false."
Give me a break.
> I don't think this is really a reasonable proposition as a general matter.
Then I won't use your software. I will recommend against it. I will ensure that everyone understands it's faults. That's the only reasonable response as an irritated user.
How is it an unsupported way of configuring Firefox when they literaly code the browser to accept and parse the config file and set the settings?
A supported way would imply guarantees of compatibility and stability.
These sort of perfs in user.js or about:config are clearly labeled as not having those guarantees and therefore "unsupported".
As someone who extensively customizes the look, feel and way that firefox works I think this makes sense. I of course would like mozilla to make a lot more options "officially" supported, especially turning off anything that contacts a remote server without explicitly wanted.
It's boggling my mind that FireFox is 2% of the browser market now. I'm hoping that includes mobile where it makes a lot more sense. But even so, really sad. It is actually a great browser (putting aside this issue of AI features being pushed) and I actually do have much greater trust that it is respecting my privacy over Chrome and Edge.
I'm not sure if I ever used any LLM features from Firefox besides exploring, but so far it seems great they're trying local approach instead of sending it to some "partner".
In the comment section here I see lot of people complaining about the fact it's enabled by default as well as some concerns about resource usage. Could someone experienced in desktop app architecture explain if disabling them functionalities makes Firefox that much faster or using less resouces? I'd assume that those functionalities are kind of loaded on demand?
"AI" became such a keyword that seem to instantly give either positive or negative response, it's also an advertised feature of every second app with many of them just forcing AI into you just because of hype. This doesn't seem to be a case in Firefox - so I highly disagree with the title - the features are there but they don't go into your way if you don't want them, therefore it's easy to just use it, only when needed
As long as it's local I think it's okay.
I also think that we in the long run will probably let machines do most tedious browsing for us-- digesting ad-ridden websites, digesting interfaces. The LLM navigates the actual web, presented to maximize revenue and maximize user engagement, time spent on the website etc., but we only see actual content, carefully arranged to be as comprehensible as possible, and if we want to communicate with somebody through a website controlled by others we formulate the message and the LLM submits it.
Is it local though?
I'm so confused, they have a whole page that starts with how they run everything on device[0]. But the page ends telling you how great it is that you can choose your AI service which then leads to a page where ... all the options are hosted AI providers and half the page is about you have to read their terms of service and they all invade your privacy[1]. I stupidly enabled Gemini at some point and now there's no obvious way in the UI to disable it. I can only switch to a different hosted service provider.
I searched in settings for AI, model, LLM, nada.
Whatever good things they did here they totally screwed up the comms and UI execution of it.
[0] https://blog.mozilla.org/en/firefox/firefox-ai/ai-browser-fe...
[1] https://support.mozilla.org/en-US/kb/ai-chatbot
What is this all about? Where would I find any of these LLM features? What do they even do?
When I go to about:config and enter "browser.lm", there's nothing displayed. That's on the latest 144.0.2 release.
Is any of it somehow related to this? https://www.heise.de/en/news/One-API-for-all-Mozilla-ends-LL...
>I would think most non technical users would just use a different browser.
I would think most users would ignore the features they don't like? Idgi
I think that (optional) AI features are fine, but the model selection only shows the big providers by default. You can enable local LLM support under about:config -> browser.ml.chat.hideLocalhost (set to false). It will check localhost:8080, which is not a great choice for a fixed port. I'd prefer if I could just point it at any OpenAI-compatible endpoint. Either way, seems to work well when serving a local model with llama.cpp.
What is your favorite Firefox alternative?
Good coffee and/or long walks.
Zen revolutionized my workflow!
I've been using Vivaldi alongside Firefox anyways for the edge cases where non-chromium browsers refused to work but I've switched fully now because of the latest round of AI crap. At least Vivaldi has explicitly committed themselves to keep it out of the browser.
https://vivaldi.com/blog/keep-exploring/
Floorp is good so far switched without regrets
Sharing my configs, hopefully they'll be helpful to someone. I don't even use Firefox but I keep the settings updated (prefs.js can be modified by Mozilla/Firefox, user.js does not change).
NOTE: Some settings might block too much, edit and use as you please.
https://rentry.co/browserconfigs / https://rentry.org/browserconfigs
Wondering how these chatbot links in the sidebar are any different from bookmarks.
I've found the "Copy text from image" menu command to be useful from time to time, and the features I'm not interested in seemed easy to turn off/dismiss/ignore?
They’re opt-out and can be disabled in the settings or fully disabled in about:config. Definitely annoying but not enough to make me switch to a Chromium based browser.
> They’re opt-out and can be disabled in the settings or fully disabled in about:config.
Any setting where there's a difference between "disabled" and "fully disabled" is user hostile. And, for a company that advertises itself as all about respecting the user, Mozilla sure does love their user-hostile decisions.
That's disappointing. If they can do EME-free builds, they can do LLM-free builds. This isn't a 'tricky issue'
https://hacks.mozilla.org/2014/05/reconciling-mozillas-missi...
The built in LLM translation is great. I'm leaving that enabled. It's useful and it's a perfect application for local LLM to be better than the equivalent corporate services.
I disable a ton in default FF and even run the unbranded versions so that it's not trialware (FF branded builds all expire when their baked in add-ons CA TLS certs expire). But the LLM translation? That's finally a good feature.
It isn't clear what browser.ml.chat and browser.ml.pageAssist are associated with in terms of features. Does anyone know? I tried disabling all shown in the write-up and local LLM translation still seems to work so I assume it's something else.
The built-in translation is not LLM, it's NMT
NMT is a category containing both transformers and deep RNN. The Mozilla translation models are transformer LLM NMTs trained via Marian https://marian-nmt.github.io/ (ref: https://github.com/mozilla/translations/blob/main/docs/READM...)
>Alternatives # There are a few projects that are forks of firefox with these features removed.
>I would think most non technical users would just use a different browser.
I would think they would list one or two of them under the “alternatives” section…?
Waterfox (which I maintain), Tor Browser, Mull (based on Tor) and I assume LibreWolf as well?
Love waterfox! Thank you for your awesome work.
I am aware of several forks myself, it was just odd they listed none lol
https://www.waterfox.net/
LibreWolf seems to default most of these off. extensions.ml.enabled defaults to true, but the others look good
LibreWolf in general is a browser you need to enable things if you want them rather than have to disable spyware like for mainline firefox
Big fan of librewolf
Honestly, I haven't even noticed LLM features in the desktop version. But I found it really annoying in the mobile (iOS) app. There's a "Summarize Page" feature that's adjacent to "Find in Page." It's easy to mis-tap when you're just trying to search for a term on the page. It's also activated by the shake gesture, which can happen accidentally.
Three dots -> Settings -> Page Summaries to disable that.
Firefox for Windows now have a button “Talk to OpenAI”
Where can I find these LLM features in Firefox?
Last update has these features in side bar along with side bar tabs, confirmed on windows but couldn't see those on arch with sway wm
In your address bar type: about:config
You can search for the configuration options mentioned in the article (e.g. browser.ml.enable) in there.
Right click -> "Ask an AI chatbot"
It's important to remember that if people think of Firefox as a way to escape LLMs being pushed on them in Chrome, they might accidentally attract users from Chrome. It's the only possible reason why this wouldn't be a plugin. Why else would a user-focused browser consciously come up with strategies that attack user preferences?
You just don't understand the eleven-dimensional chess that you have to play to get from 30% marketshare to 2% marketshare. They have to think they have a winning strategy, judging by the way they talk to everybody who criticizes their decisions.
sure the smart tab group features are kind useless (but tab groups are grate)
but I don't understand why people make such a deal about it
it looks a loot like "I don't like this so it mustn't even exist even if some people like it" mentality
like
- all of this features are opt in, they don't just randomly send your data somewhere or randomly eat your CPI resources (without you having enabled them)
- the smart tab group features(1) are the only very visible and kind dump/useless one, and you can disable them without tweaking internal configs unser "Addons > On-device AI" (but it won't disable the button in the group tabs menu which is used to enable that feature in the first place)
- for many users side translations is an important must have feature, it's one of the AI features, and having a local on-device no data send to servers translation is very desirable (to be fair US users probably don't care nor realize it matters, but it really matters) (To be fair the OP didn't disable this either)
- for increasingly many users integration with chat bots (e.g. ChatGPT) is a must have browser feature(2). The integration is also only visible in a context menu(3) and hidden in the a selector for the side bar and needs to be explicitly setup before it does anything. So a feature some people want and treat as must have and now can have and for everyone else it's pretty much out of the way and not pushed onto you at all. So why insists that it's not okay to have it even in a selector hardly any power users will ever use (because the other options there are bookmarks, and history, both with well known widely used keyboard shortcuts)
- the preview was a "Labs" (i.e. experimental preview) feature you have to explicitly enable and then use the right hot key to use, and it's one of this "some people expect this feature in 2025" features (but not sure what happened to it, it's neither in Labs nor enabled in any FF browser I have, so maybe they discontinued it)
(1): They IMHO are pretty dump and seem more like a way to setup the whole infrastructure around them with a trivial to implement feature then a in depth feature mozilla cares about. I guess.
(2): On HN it often seems like only enthusiasts and not many non-tech people use ChatGPT and co. _That is not true at all_. When it comes to "simple" ChatGPT usage far more non technical users use it then technical one. Like for dump stuff like shopping list, holiday check lists, etc. The problem OpenAI has isn't that they don't have a lot of users, it's that they need a absurd amount of high paying users. But the tasks most non-tech people use generic ChatGPT for aren't worth a lot. Like even the Plus plan is a "only if you are wealthy enough to burn 20 bucks every month without caring" option in such a context and the Pro plan just absurd.
(3): Is anyone except non technical users still using browsers context menus in a way where a "Ask AI Chat-Bot" entry matters???? Like copy, past, undo, redo, select all have trivial short cuts. Language/spell check is rarely touched and inspect has a short cut where you press Q after right clicking.
deleated from most recent linux install on a new to me desktop. vile I just want to work(create stuff), there is nothing that any "service" can do for me. the relentles attempts to get in the way of sending and recieving information is mindbogling, and I am going to be getting an analog telephone adapter, so that I can start useing fax, over cellular data, and making paper plans ,taking pictures, and converting to pdf's to avoid the wasted time and cost of software solutions the surreal part is that the LLM's have found me, and are recomending me to people looking for custom metal, so much for a low profile there, and somehow my number got out on indeed, and I got a wave of people calling, applying for "the job" but bitterly complaining that the app they downloaded as part of that, wasn't working right. fine, ok, whatever.
but not on my machine