I don't think we should equate Firefox's AI plans with other ones.
Firefox's AI will be trained entirely locally with data that you choose to give it, and won't send information back to Firefox.
By the sounds of it it won't be a chatbot either, but rather an aid for finding more sources, pointing out fake reviews, assisting in (offline, local!) translation, etc.
My two main issues with AI are unethically sourced training data, and hoovering up personal information when you use it. Neither are a problem for Mozilla's AI plans. This is how AI should be done.
Well this is the thing here. We're in an era of time in technology where we DON'T want people having as much of our data. Whether it's for good or bad use, we just don't want it. How hard is that for these companies to comprehend?
The internet was fine without this sort of thing. We were fine without AI. Why complicate it at all?
Yes, and expensive. This is often overlooked when people just say don't worry it's FOSS. The enshitification happens slowly, and by the time there is outrage about how bad it becomes the last non intrusive fork might be several years old and take even more work to modernize. I'm not giving up all hope, but you are correct, it would be a very ambitious community undertaking to keep such a thing competitive with the plethora of evil browsers out there.
Vivaldi, although it's extremely nice tab management system and unique features, isn't open source. But probably the better chromium browser of the owns you mentioned.
There are a plethora of sources out there of good reasons not to use these browsers. But seriously, Firefox is an excellent browser, treat yourself to better privacy 🦊