TL;DR: Mozilla has a new CEO and a new mission: transform Firefox into an AI browser. That has run into some snags, as Firefox users don’t seem that interested in AI. Mozilla is forging ahead, utilizing deceptive patterns (previously known as dark patterns) to nag and annoy people into enabling AI features. You can see this in the introduction of Link Previews, an extremely invasive anti-feature that exists solely to push AI into your experience.

all 14 comments

sorted by: hot top controversial new old
[–] 11 points 10 months ago (9 children)

Is this guy for real?

Mozilla says that key points are processed locally to protect your privacy in the release notes, but says nothing about leaking your privacy in showing the link preview (and enabling it by default).

As opposed to the case where you don't have a link preview, and you click on a website to see what it contains, and they get your IP. The author seems to think Mozilla should have protected our privacy by having someone act as the proxy for the request. Because involving a thirds party that receives all these requests and does work for us for free is absolutely how we protect our privacy.

The user might also have mobility impairments that makes a fast click harder, resulting in a longer hold time.

Yes, a feature clearly designed for pushing onto that juicy "people with mobility impairments" userbase.

I don't like the direction Firefox seems to be headed in, but damn people really enjoy getting outraged over everything they do. Around here they get ten times more shit than any other comparable project.

  • source
  • hideshow 9 child comments
  • [–] 5 points 10 months ago

    The author seems to be more interested in generating outrage than anything, but I think the point about AI still stands. From a UX standpoint, key points that may be incorrect are a terrible idea. That they originally intended to force AI on the user, at least from how it seems, is problematic.

    The author's privacy and accessibility concerns seem artifical to me.

  • source
  • parent
  • [–] [S] 5 points 10 months ago (2 children)

    As opposed to the case where you don't have a link preview, and you click on a website to see what it contains, and they get your IP. The author seems to think Mozilla should have protected our privacy by having someone act as the proxy for the request. Because involving a thirds party that receives all these requests and does work for us for free is absolutely how we protect our privacy.

    But that is exactly what Mozilla is telling us -- trust us.

    Why was the feature added if my browser is going to browse to the page anyway? What is the value add? I was looking for some way for it to make sense - ah right, it could be a privacy preserving feature - I can preview the link and verify whether I want to visit it before I actually visit it. But that isn't how it works.

    Yes, a feature clearly designed for pushing onto that juicy "people with mobility impairments" userbase.

    Love that you ignore all of the people who are currently seeing the popups and not understanding why.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 10 months ago (1 child)

    Why was the feature added if my browser is going to browse to the page anyway? [...] it could be a privacy preserving feature.

    It's just supposed to save you time and effort.

    If anyone has real concerns about having their IP leaked they should be using a VPN (I think Proton has a fairly generous free tier) or TOR. Relying on a link preview feature like that would be like wearing a condom against the rain. It will technically increase your protection, but you will still be really quite exposed.

    Love that you ignore all of the people who are currently seeing the popups and not understanding why.

    No, I just took his objection at face value.

  • source
  • parent
  • hideshow 1 child comment
  • [–] [S] 1 point 10 months ago*

    I added a section to my post with some additional comment.

    I began thinking of privacy because Mozilla was clearly thinking of it when designing this feature, but I don't think they really thought it through.

    People's browsers are visiting pages that they never intended to. If a random extension did that, you would say that it was violating your privacy. The browser does it, and you get people defending it as "optional". Yes, but the user never installed the malware extension that is leaking your privacy. It is your browser doing it in an automated update.

    If you don't think this is a privacy issue, why doesn't the next version of Firefox just visit every page on every page that I visit, so that when I hover over a link, I can get a link preview immediately, without needing to wait. That would save me some real time and effort!

  • source
  • parent
  • [–] 2 points 9 months ago* (last edited 9 months ago) (3 children)

    The author seems to think Mozilla should have protected our privacy by having someone act as the proxy for the request.

    On the proxy part, they actually already have that and using it for some other parts:

    https://support.mozilla.org/en-US/kb/ohttp-explained

    TL;DR: Imagine an HTTPS-over-HTTPS proxy. Try to explain it like something groundbreaking without referencing existing tech. Now you have OHTTP.

    https://firefox-source-docs.mozilla.org/browser/components/mozcachedohttp/docs/index.html

    https://www.fastly.com/blog/firefox-fastly-take-another-step-toward-security-upgrade

    It makes me scratch my head a bit why I've never see it enabled for DNS-over-HTTP in default stock Firefox config despite it being supported for years - the endpoints are just not configured. You have to know about it and configure the barely documented URL in about:config for that. Unlike for newtabpage and the FF shopping feature where OHTTP is used by default. Infra costs?

  • source
  • parent
  • hideshow 3 child comments
  • [–] [S] 2 points 9 months ago* (last edited 9 months ago) (2 children)

    Interestingly, I just interviewed the Waterfox developer, who actually references Oblivious HTTP and his interest in developing this into a paid feature for Waterfox.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 9 months ago (1 child)

    That is interesting!

    BTW in case you're not aware, direct links to fedia.io like the one you posted just lead to a loginwall so you probably don't want to share those publicly. This one via beehaw.org works for everyone, though: https://beehaw.org/post/24563411

  • source
  • parent
  • hideshow 1 child comment
  • [–] 9 points 10 months ago (3 children)

    Zen figured out link previews without using AI and the solution is really as simple as it gets. Maybe stop trying to manufacture problems for AI to solve?

  • source
  • hideshow 3 child comments
  • [–] 5 points 10 months ago (2 children)

    From the linked article I learned that Firefox's solution also doesn't use AI, not by default at least.

    And the Zen way of doing it has the exact same (imaginary) privacy issue for which the article blames Firefox.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 10 months ago (1 child)

    The feature was introduced as a way for users to get relevant information faster, by providing them with an image, the webpage title, and AI-generated key points.

    The AI part was made optional. That doesn't mean they didn't try.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 4 points 10 months ago

    They create an AI feature, they realise people don't want it, and realise a minimal one they can turn on for everyone in a thin-end-of-the-wedge approach.

    OR

    They create a feature with AI, realise it's controversial, so they figure out a minimal version, they split the parts with and without AI, and enable the non-controversial one by default.

    The facts are the same, just a different narrative. Which is legitimate. Realizing that's what it is is non optional.

  • source
  • parent
  • 13 comments in c/technology@lemmy.world view that post
    [–] 20 points 10 months ago

    I'm really grateful for the introduction to deceptive patterns here.
    I was not aware of it, and I think it's important to have language that can describe specifically how tech companies are trying to coerce people.

  • source
  • [–] 23 points 10 months ago

    Why won't these pesky peasants shut up and give us our data?

    Who could have known that the people who voluntarily use a tiny niche browser mostly for privacy/idealism reasons wouldn't wany just another AI browser?

    Honestly, how out of touch with their user base can they be?

    If I want a good browser that works with every website, I'd take something chromium based. If I want an AI browser, I'd take one of the many chromium based ones.

  • source
  • [–] 37 points 10 months ago (4 children)

    Not this seems overly reachy Firefox hate. This feature pushed months ago didn’t it? I managed to turn it off without even having to dig, and I know this because I don’t even remember how to do it.

    Firefox defaulting a link preview gesture to on, but making the AI off by default, isn’t a dark pattern. Every new update doesn’t turn this back on if previously disabled. And trying to twist this into “Firefox is using dark patterns to trick you into using AI” just feels like an attempt to shove “Firefox bad” down people’s throats.

    I find the amount of Firefox hate Lemmy seems to display so odd, especially when literally every other browser out there, from chrome to brave to opera, and everything in between, barely ever gets even a mention.

    This stinks of negative advertising, and I’m getting so fucking tired of it.

  • source
  • hideshow 4 child comments
  • [–] 6 points 10 months ago (1 child)

    I tried the link preview feature as well, and to say the response to it is overblown is putting it mildly. I haven't looked at the source code, but based on how it appears to work I'm not sure it even qualifies as AI. It basically selects 2-3 sentences from the reading mode version of an article, but the selection is so bad it might as well be random. Not surprising as it's a tiny model that runs locally and is only given a second to make the selection.

    I actually laughed when I saw it - this is what all the weeks of fuss were about?

  • source
  • parent
  • hideshow 1 child comment
  • [–] 17 points 10 months ago
    [–] 10 points 10 months ago (3 children)

    Its open source right?

    We can just make a “dumbfox” browser without ai

  • source
  • hideshow 3 child comments
  • [–] 5 points 10 months ago

    I've switched to IronFox on mobile and Waterfox on computers, personally.

  • source
  • 5 comments in c/linux@programming.dev view that post
    [–] 6 points 10 months ago

    I have been specifically thinking about how big tech keeps manufacturing consent for AI, having it on by default and counting accidental undesired uses for "95% of people adopted our AI features!" and all this… glad to see an article that has a title reflecting my thoughts. Looking up manufactured consent for AI usually brings things on how people can use AI to manufacture consent or change narratives, not on "consent to AI" being manufactured itself.

  • source
  • [–] 5 points 10 months ago (2 children)

    Wait, is "dark pattern" taboo now?

  • source
  • hideshow 2 child comments
  • [–] 4 points 10 months ago

    For me, as long as Debian still packages it and disables these features, I’ll be fine, but LibreWolf looks more and more tempting these days, and having tried it a bit, I can live with the minor annoyances.

  • source
  • 2 comments in c/linux@lemmy.world view that post
    [–] 4 points 10 months ago (1 child)

    I honestly think link previews are a good idea and can be very beneficial for users if done right. It would be good to know if a site could generally be trusted before I click a link. It‘s just hard to do that right, though.

  • source
  • hideshow 1 child comment