@firefox, the “last privacy-respecting browser” now nags you with in-browser pop-ups to let AI generate “key points” when you long-press links.
Mozilla CEO: “AI should always be a choice – something people can easily turn off”
Then why the fuck is it off by default? Why the fuck am I getting pop-ups asking me to try features I didn’t ask for? That’s not a choice. That’s opt-out with a fucking marketing budget.
What the fuck happened to you, Mozilla.
They spent WEEKS in damage control promising an “AI kill switch” and then shipped it fucking disabled. That is the most gaslit UX I have ever seen in my life.
“Help me @librewolf - you’re my only hope.”
Settings > AI Controls > Block AI enhancements. Do it now, because they won’t do it for you.
—
Edit: a few corrections thanks to @Feyd:
“Hover over links”
- It’s a long-press / context menu action, not a passive hover. That’s a meaningful distinction because hover implies it’s happening constantly without intent, which is way more invasive than what’s actually happening.
“Sending page content to ML models”
- the default link preview (before you enable key points) just reads the Open Graph meta tags – the same og:title / og:description metadata that generates link cards in Slack, Discord, iMessage, etc. That’s not AI, that’s just HTML parsing.
- even when you DO opt into the AI key points feature, it runs a local on-device model, not shipping your page content off to some cloud endpoint.
@k3ym0 @firefox @librewolf @Feyd - currently using Waterfox, which *appears* to have none of this nonsense attached.
I agree with not wanting link previews, but there is some misinformation here:
when you hover over links
It is a 1 second hold or a context menu option.
is sending page content to ML models
Unless you’ve specifically enabled the key points feature, it just shows the open graph tag content (https://ogp.me/). This is also what is used to display a card when you put a link in a chat app.
If you do enable the key points feature, it is not sending page content over the network like you imply.
It uses a local model and it doesn’t even download that model until the first time you ask for the key points.
I encourage you to keep complaining about things you would like you be different, but you should really keep your complaints factual, because you weaken your points when you mix them with misinformation.
https://support.mozilla.org/en-US/kb/use-link-previews-firefox
@Feyd Fair corrections, thank you. The hover thing was sloppy on my part – it’s a long-press/context menu, not passive. And the key points feature does run a local model, not cloud inference. I’ll edit the post.
The default-on nag UX is still bullshit but you’re right that mixing valid complaints with inaccurate claims just gives people a reason to dismiss the whole thing.
What nag are you talking about?
- I haven’t seen any sort of nag that occurs before activating a link preview
- After activating a link preview there is a cog button that takes you directly to the settings where you can disable link previews entirely, in case you used it by accident or tried and decided you didn’t like it.
- The key points part of the UI has a toggle to hide it, and it stays hidden for every subsequent link preview, even through restarts of the browser.
They spent WEEKS in damage control promising an “AI kill switch” and then shipped it fucking disabled. That is the most gaslit UX I have ever seen in my life.
Okay, so I hate these features too, but the “Block AI enhancements” switch does work exactly as advertised. Should it be disabled by default? IMO yes, but this is not “gaslighting”. Its existence is honestly the only thing keeping me from dropping Firefox entirely.
I get that devs want people to discover and use the features they’re worked on, but I don’t understand how Mozilla can be so willfully ignorant of what their user-base wants. It’s not like we’re quiet about it.
after the update it literally pops up and tells you its part of the program now and gives you the option to turn it on before doing anything else.
Some people just want to be angry.
That is the most gaslit UX I have ever seen in my life.
That is the most obnoxiously incorrect therapy speak usage of “gaslit/gaslighting” I have ever seen in my life.
@TheTechnician27 you’re technically making a point about language inflation, but you picked maybe the worst possible example to die on that hill because “we said it’s opt-in but it’s actually opt-out” is genuinely closer to gaslighting than 95% of the times people use that word online.
ahem
tap mic
align notes
adjust glasses
Lying is not the same thing as gaslighting, you colossal fuckmuppet.
if you’re going to play grammar Nazi with someone else’s imperfect English, at least exhibit some basic literacy yourself and read what they said.
@k3ym0@infosec.exchange said they were closer, not that they were right.
@TheTechnician27 “colossal fuckmuppet” is doing a LOT of heavy lifting for someone claiming the moral high ground on precise language usage.
Oh, no, this follow-up comment reinforces that I got it exactly right.
Deleted by author
if you’re going to correct someone else’s imperfect English
It’s not about “imperfect English”; it’s about lazily, selfishly abusing real therapy terms in places they don’t apply in a way that strips away crucial language from victims of abuse and genercizes it to absolute meaninglessness. I’m not going to argue the truth or merits of “closer” when they’re a fucking million miles away from what the word means; you’re upset with me for not playing into their fallacious relative privation bullshit, and I categorically don’t care.
@firefox @librewolf OH MY GOD.
The Firefox mastodon account automatically boosts your post when you tag them.
lmaooooo
@k3ym0 @firefox @librewolf that account isn’t affiliated with Mozilla at all. They used to operate an official account on their own mozilla.social instance, but shut that down and as far as I’m aware they aren’t on the fediverse at all anymore (their website links to a bluesky and a few other social platforms, but no fediverse).
