• Clent ( Clent@lemmy.dbzer0.com ) 
    link
    fedilink
    English
    arrow-up
    9
    ·
    3 days ago

    It’s a question of liability.

    OpenAI wants zero liability.

    But clearly the answer is a non-zero value. I don’t see how they aren’t 100% liable when it’s their employees who execute the process.

    They are starting and running their own machinery. The industry likes to pretend software is some intangible but as hardware prices have demonstrated, software is very much physical. And it’s not narrative, it’s instructive.

    • Hirom ( Hirom@beehaw.org ) 
      link
      fedilink
      English
      arrow-up
      2
      ·
      3 days ago

      Large AI companies know their tools are dangerous, and are shouting it from rooftop.

      Knowing that, they keep developing them, and ran agents with little supervision, insufficient sandboxing, and ended up attacking third parties.

      I hope there’s liability.

      • Clent ( Clent@lemmy.dbzer0.com ) 
        link
        fedilink
        English
        arrow-up
        1
        ·
        2 days ago

        If everything you said is provable, they’re opening themselves up to criminal.

        It’s not clear they can lobby their way out of this without empowering everyone.

        These models are math. Coding is solved. Code is math. The models are solved.

        If they can hack others, then we can hack others. We can hack them. There is an inevitability to this but for the moment we are still governed to laws. And laws have to be applied evenly or not at all, according to the law.

        Law themselves being part of a poorly defined math system.