On 16 July, the AI model repository HuggingFace posted a security incident report. Well, I say “incident report” — it’s written like an AI doomsday press release: [GitHub] Earlier this week, we det…
  • EFreethought@awful.systemsEnglish
    6·
    1 day ago

    So far, all of these “model breaches containment” stories have been fake.

    But if one actually did, than that would be a signal to drop AI and nuke data centers from orbit.

    This is not the genius-level marketing OpenAI thinks it is.

    • scruiser@awful.systemsEnglish
      4·
      19 hours ago

      that would be a signal to drop AI and nuke data centers from orbit

      On Lesswrong I’ve seen some posts contemplating why this “warning shot” isn’t making any political waves. Also, in response to the fact the OpenAI has released basically no details, I read one poster realizing that actually they need external governance and regulation on the AI companies.

    • lurker@awful.systemsEnglish
      4·
      22 hours ago

      Its truly astounding that they put a marketing spin on committing a crime but thats the AI industry for ya