After a seemingly endless stream of tone deaf news from Mozilla, we’ve finally got some good news for Firefox users. As the company’s been hinting at for a while on social media now, they’ve added an “AI” kill switch to the latest Firefox nightly release, as well as a set of toggles to disable specific “AI” features.
You can choose to use some of these and not others. If you don’t want to use AI features from Firefox at all, you can turn on the Block AI enhancements toggle. When it’s toggled on, you won’t see pop-ups or reminders to use existing or upcoming AI features.
Once you set your AI preferences in Firefox, they stay in place across updates. You can also change them whenever you want.
↫ Ajit Varma at the Mozilla blog
I’m particularly enamoured with the specific mention that the setting will remain unaffected by updates. It’s incredibly sad that Mozilla even has to mention this, but they have nobody to blame but themselves for that one. None of this is enough to draw me away from Librewolf and back to Firefox, but at least it gives those of us who prefer to keep using Firefox the option to disable all of this “AI” nonsense. Also, there’s no Librewolf for POWER9, so I have to use Firefox somewhere.
It’s unlikely Chrome or Safari will get such clear “AI” kill switches, so it might become a reason for some to switch to Firefox from Chrome or Safari.
I’m glad they are finally stepping up and doing what they promised, though ideally they should ship a browser completely free of any “AI” nonsense. Actually they do, but only for mobile: Firefox Focus is about 99% of what most Firefox users need in a reliable, safe, private, and performant browser, but they won’t release it for desktop OSes and they won’t tweak it for use as one’s main browser on mobile, rather they position it as “private tab mode plus”.
Same, OpenBSD does not have a port for Librewolf, but their Firefox and Firefox-ESR ports are already most of the way there. It also helps that Firefox on OpenBSD is properly sandboxed with pledge(2) and unveil(2).
So its not off by default, and can be turned on magically by next update! I really dont trust these liars anymore!
IMO this is how it’s going to look like: you disable AI features, restart browser, make an update and boom! AI is silently creeping back. Not necessarily because they do it on purpose to users who disabled it, but because it’s OPT-IN, so enabled by default. Thus, they ASSUME users are using it. This assumption alone will cause things to go haywire because they will code everything with AI at the center, make lots of silent assumptions and depencencies.
It’s indeed very sad, but Mozilla and pure Firefox are quite over for private open source enthusiast. Forks? will work just fine untill Mozilla makes way too much assumptions about users using AI and it will be too hard to cut that crap out of the code without breaking things.
If the biggest selling-point of what you’re trying to push is the off switch… Maybe that isn’t what you should be trying to push.
Thom Holwerda,
Hopefully I’m not being too direct but I think it’s worth having a discussion about the way osnews covers AI news. I think osnews should differentiate between different types of AI. It’s a large umbrella term in computer science covering vastly different technologies. Lumping them all together does not do them justice and IMHO osnews has been guilty of doing that. I understand your personal reasons for holding a grudge. I feel that when all is said and done the impact on jobs will be even more widespread than people realize. However as a reader it seems you are holding a grudge against all AI technology without enough nuance. In short I’m trying to see if I could persuade you to be more selectively critical of AI technology using more objective criteria (ie harms jobs, privacy, etc) rather than treating the entirety of AI as a generic foe.
That’s on the industry no? “AI” is a marketing term, I don’t think it’s on Thom to come up with differentiations. If a company doesn’t want to get lumped in with the others, they can do that marketing work themselves.
paisley,
No actually. AI was used in computer science many decades ago and not at all as a “marketing term”. According to wikipedia “Artificial intelligence was founded as an academic discipline in 1956″…
https://en.wikipedia.org/wiki/Artificial_intelligence
Given the technical audience here, I don’t think I’d be the only one to appreciate a technological focus. I’m not against covering how corporations are marketing or abusing AI, but I don’t think that deserves to become the entire scope of AI either.
> No actually. AI was used in computer science many decades ago and not at all as a “marketing term”.
Sure it was, so what? Marketing has taken over, everything is AI now. From LLMs to good old machine learning. I know of no technical news site that differentiates anymore, why should here be different?
Serafean,
That’s not marketing, these actually do belong under the AI umbrella, The fact that very different technologies exist under it was the whole point.
I have to agree with paisley and Serafean on this. The term “AI” has lost most of its original meaning as it is currently used by the industry to mean anything automated by a computer. Based on the current definition popularized by the industry, a script written by a human to automatically filter a list or whatever, would qualify as “AI” because the computer is performing a normally manual task. That’s insane and nonsensical, but ask the average non-technical person if a computer automating a menial task is considered “AI” and they would respond that it probably is.
The burden is on the industry to differentiate the various technologies under the “AI” umbrella. And to be fair to Thom, he has many times in the past called out the differences between the various technologies under the “AI” label, and was attacked for that as well. It’s a “damned if you do, damned if you don’t” situation and I feel you are unfairly criticizing him over it.
Morgan,
I don’t believe that I come across that usage often, but to the extent that it happens, wouldn’t it still be important to differentiate the good from bad? How do you justify throwing shade at everything under the AI umbrella without regards to anything else?
I would absolutely agree with Thom that there are bad actors, but the narrative on osnews seems to be that AI itself is bad rather than specific applications of AI. Given that AI refers to very different technologies solving problems through different practical and ethical approaches, I find non-specific blanket criticism hard to square away.
The onus is also on those covering it too. Anyone covering any topic can choose whether to gloss it over or be a stickler for details. Whenever something becomes mainstream, it necessarily brings down the average understanding of what is otherwise a specialized discipline. So of course there’s going to be a lot of simplification in mainstream media, it’s what they do because of their audience.. But it doesn’t follow that those of us in more technical circles need the topic to be watered down to the public’s level.
On the one hand I feel out of place making any requests to Thom, it’s his site and not mine. I don’t get a vote and I don’t want to come across as a self entitled jerk haha. But on the other hand I don’t feel it unfair for me to try and persuade Thom that AI is too broad to meaningfully vilify AI so generically. I learn a lot here and I am not above being educated about why some company’s AI offerings are harming our ethical norms. But boiling things down to “AI=bad” without more specific justifications buries the ethics behind too much hand-waving.
I want to explicitly say we should be critical on AI topics, but that criticism needs nuance. For example there’s a big difference between AI technology designed to run locally versus AI designed to capture our data and run under someone else’s control.
https://support.mozilla.org/en-US/kb/on-device-models
To me such distinctions are ethically extremely important. Glossing over these details like this does an injustice to those trying to make AI better and more respectful of our rights. I would like for privacy respecting AI to have a competitive advantage over privacy intruding AI, as I think everyone should. But when the media won’t even bother to distinguish between these and faults them all the same, we’ll end up with the worse AI winning.
@Alfman:
I hope you and I can agree that the current corporate “AI” bubble does absolutely nothing to advance the state of the art, and is largely a money grab and environmentally harmful phenomenon. That is what Thom is calling out, and rightfully so. He has mentioned in the past that locally run “AI” is intriguing and a good use of the technology, but as I said (and you seemed to ignore) he was vilified as a “hypocrite” for it, so why should he bother differentiating after that kind of treatment from the readers here? It’s not that I don’t understand your position, it’s that, at least here on this site, you probably won’t see that kind of separation of “good AI” and “bad AI” reporting for the foreseeable future, because there’s no point.
Morgan.
AI is the latest gold rush, but it doesn’t change the way I see corporations. Nearly everything corporations do is largely a money grab. And indeed it is also true they don’t care about their environmental impacts.
There’s no specificity in anything being called out though. My point isn’t to say that people with anti-AI opinions are wrong, however it is sensible to ask exactly where the ethical problems lie and how it applies on a case by case basis.
I can’t comment on that because I don’t know what you are referring to, but maybe a link would refresh my memory.
I don’t think it invalidates my points, but you are right that there’s a decent probability my request will be ignored. Still, I think it’s better to ask.
I care about osnews and I believe that fleshing out the ethical arguments could make AI articles better. Rather than saying something is bad, explain exactly why it’s bad and what could be done to make it good, etc. When someone does something good with AI, highlight it! Blanketing the field with generic anti-AI arguments muffles the advantages of those trying to make AI progressive and leaves the bad actors to win by default.
Let’s use an example. It’s far more beneficial to Microsoft executives that that AI critics do NOT distinguish between privacy respecting AI and data hording AI. The less aware customers are about such distinctions, the better it is for microsoft. Osnews hasn’t even covered any of the open models that could help provide some degree of competition for microsoft. The lack of media coverage for AI that offers owner control and privacy is a boon to the AI juggernauts who want to control it all.
@Alfman:
If there’s anything I’ve learned from our previous conversations here, it’s that when we both start repeating ourselves, it’s time to bow out. Peace brother.
> I don’t think I’d be the only one to appreciate a technological focus.
You are not the only one! I would very much agree with you asking for a more objective less biased few on AI as well as on Corporates.
In this particular case of “Firefox AI” I even agree with Thom but in general he appears to be very bitter on both topics and this harms the site.