• BrianTheeBiscuiteer@lemmy.world
    link
    fedilink
    arrow-up
    9
    ·
    3 hours ago

    Training isn’t my biggest concern, it’s the data collection beforehand and the use of the resulting model. AI companies need to be held liable for rogue AI agents. Anthropic supposedly hacked HuggingFace due to rogue AI and I’m 90% sure they’re not being investigated in any way for computer crimes.

    • 2ndDegreeOWO@lemmy.world
      link
      fedilink
      arrow-up
      5
      ·
      3 hours ago

      We also need to do something about the DDOS-level threat AI content crawlers pose to almost every website on the internet.

      Especially since a lot of websites forbid it in their TOS, but AI companies just do it anyway.

  • amniotic druid@lemmy.world
    link
    fedilink
    English
    arrow-up
    9
    ·
    3 hours ago

    Dems 🤝 not knowing wtf theyre talking about

    Bolster the SEC, investigate NVIDIAs circular financing, eliminate corporate welfare and the whole thing folds anyway.

    Fearmonger about a sci-fi concept, you’re nothing but a marketer

  • A_A@lemmy.world
    link
    fedilink
    arrow-up
    1
    arrow-down
    1
    ·
    54 minutes ago

    … permission to train AI in USA ; has no effects in China, so no global effect.

  • eicker@lemmy.worldOP
    link
    fedilink
    arrow-up
    17
    arrow-down
    3
    ·
    4 hours ago

    AI absolutely needs guardrails. But letting a new agency redefine »advanced AI« whenever it likes could turn regulation into a very expensive velvet rope. Google and OpenAI get compliance departments; smaller labs get a number and a chair at the DMV. Nothing says competition like needing government permission to compete.

  • Nola@piefed.zip
    link
    fedilink
    English
    arrow-up
    8
    arrow-down
    1
    ·
    4 hours ago

    I need to see more details of this

    On the one hand the industry needs more regulation and frankly, far more restrictions then is currently happening, and this coming from the progressive wing and not centrists is at least a somewhat better sign.

    On the other, most Dems are still corporatist ghouls, and I think even Bernie has shown some bad instincts on handling AI at times, and this reads eerily similar to the cartelification the AI companies are reported to themselves be seeking.

    Where instead of the Chinese approach of open sourcing the tech more, America, in the name of national and domestic security, essentially grants the existing companies defacto protectionism and in a way that will make it impossible to dislodge them.