It’s an unintended side effect that the majority of custom ROM users will appreciate.

Firefox being the first and only browser on Android enforcing Google play integrity verification wasn’t on my bingo card

  • XLE@piefed.social
    link
    fedilink
    English
    arrow-up
    33
    ·
    1 day ago

    Mozilla: insists AI is integral to Firefox now
    Mozilla: locks AI behind Google Play verification

    Truly the last great competitor to Google…

      • XLE@piefed.social
        link
        fedilink
        English
        arrow-up
        3
        ·
        edit-2
        9 hours ago

        That is totally unimportant because Mozilla still maintains total control over that source. Same story as people who say Chromium is open-source as if that’d keep Manifest V2 extensions functional.

    • atopi@piefed.blahaj.zone
      link
      fedilink
      English
      arrow-up
      2
      ·
      17 hours ago

      those 2 things could be true at once. People who use degoogled android are likely much more knowledgeable on tech and not need features normal people would need

      they arent, but they theoretically could be

  • ColeSloth@discuss.tchncs.de
    link
    fedilink
    arrow-up
    15
    arrow-down
    1
    ·
    edit-2
    1 day ago

    So to anyone who doesn’t know the venture capitalist investment side of AI: All the AI queries from people are kind of expensive as hell. Right now, even with most of the paid for use AI subscriptions many companies have paid for they’re operating at well over an 80% loss. AI is expensive. The game plan has been to keep dumping money into the very net negative revenue stream of AI long enough in order to get corporations, doctors clinics, law offices, software companies, etc so imbedded into using AI that they’ll be forced into absorbing the costs, even at a 10 fold price increase.

    That’s the plan. Increase the price by 10 fold over the course of the next several years. It will also push out any smaller businesses from being able to afford to use AI, and stop regular people from having free or low cost AI. It will all be consolidated and expensive, and prevent any new startups in industries that would benefit from AI to afford to get started.

    A cess pool.

    • Zetta@mander.xyz
      link
      fedilink
      arrow-up
      2
      ·
      3 hours ago

      I don’t think that’s the plan and it won’t work if it is, believe it or not I think these corporate soul suckers know that.

      Chinese models are literally 75x cheaper than frointer usa models with 85% of the capabilities. Chinese’s models have continually been 3 - 6 months behind frointer models. This will continue to be the case and since Chinese models are generally open anyone with a big server can run them and sell tokens, it’s gonna stay cheap.

      I’m pretty sure most LLM providers are making enough money to profit on selling infrence. They only are in the negative from massive spending on hiring the best talent and research/development into new models and the associated hardware/data center build outs to train those new models.

      The truth in my opinion is these companies are confident that if they spend enough they will reach a critical threshold where llms are actually intelligent and useful, and that is what they are betting on. Being the first company to make real “Ai” that can self improve.

      I’m sure OpenAI and Anthropic will try to charge more because they are soul sucking capitalists but that will only be viable as long as they are sufficiently in the lead from open weight models. I think Anthroptic has already 2x api pricing for their newest hype beast end of the world its so scary advanced Fable 5 model.

  • Gladaed@feddit.org
    link
    fedilink
    arrow-up
    3
    arrow-down
    2
    ·
    2 days ago

    I don’t know. This just makes sense. AI can be a major contributor to cost so making sure you generate revenue from the user just seems like the smart move.

    • Wispy2891@lemmy.worldOP
      link
      fedilink
      arrow-up
      7
      ·
      2 days ago

      If someone really wants to use their servers to do ai shit, they could spoof being a Linux or Windows client instead of the android one. There’s no difference at all.

      I’m assuming Linux, windows and MacOS builds verify the signature to make sure the executable isn’t tampered. The same can be done on Android without needing to use draconian verification apis

      • Gladaed@feddit.org
        link
        fedilink
        arrow-up
        4
        ·
        2 days ago

        An ankle high barrier is enough to reduce traffic to a point where it is no longer an issue.

        There are loads of phones and non technical users.

        • Wispy2891@lemmy.worldOP
          link
          fedilink
          arrow-up
          5
          ·
          1 day ago

          there are literally zero persons in the world that:

          1. somehow, they really need to use the mozilla ai services api for reasons (??? classify a website for tab grouping?), but don’t want to pay
          2. they are skilled enough to recompile/patch/crack the android executable and change it to do their purpose
          3. somehow, they don’t have the skills to do the same in the linux/windows/macos executable
          4. are willing to waste days in reverse engineering efforts and perpetually play a cat&mouse game to save $1/month in inference costs