More shady shenanigans to come. It was being used to allow the model to move around to hack into systems. To see how far it can go. A Red team test. Red teams ran the tests after Anthropic released Mythos early to major corporations and the government. Red-team exercises typically involve an organization testing its defenses by simulating an attack. In this case, the NSA reportedly used Mythos itself as the tool to probe its classified networks for weaknesses.

  • CameronDev@programming.dev
    link
    fedilink
    arrow-up
    31
    ·
    1 day ago

    “Chief claims” - without evidence, this means absolutely nothing. Especially as this was a higher up, who likely heard about something happening through multiple layers of reports.

      • ZDL@lazysoci.al
        link
        fedilink
        arrow-up
        16
        arrow-down
        1
        ·
        1 day ago

        The key word there is “believe”.

        Until there’s evidence, that’s all you have: a belief.

        • Anonymous_Leaker@lemmy.worldOP
          link
          fedilink
          arrow-up
          6
          arrow-down
          13
          ·
          edit-2
          1 day ago

          Ai will become more dangerous. I am telling you, it is not just about believing. It is and it will continue. It will mess up and make people crash cars for example.

          • ZDL@lazysoci.al
            link
            fedilink
            arrow-up
            14
            arrow-down
            1
            ·
            1 day ago

            The only danger this round of AI will bear is fiscal. When, not if, the bubble bursts, one country in particular is going to have a really bad year economically speaking.

            The bullshit market hype that Anthropic is putting out about how “dangerous” its “latest model” is … well, it’s exactly that. Market hype.

            • Anonymous_Leaker@lemmy.worldOP
              link
              fedilink
              arrow-up
              3
              arrow-down
              10
              ·
              edit-2
              1 day ago

              It isn’t hype if some ai is already causing problems. Examples: Deepfakes, misinformation and cyber security threats. A model was already yanked from the customers.

              • ZDL@lazysoci.al
                link
                fedilink
                arrow-up
                8
                ·
                1 day ago

                Please, just stop fucking fluffing the techbrodudes, dude. Or carry on. Won’t make a difference to me; I won’t be seeing any of it anymore.

                  • cecilkorik@piefed.ca
                    link
                    fedilink
                    English
                    arrow-up
                    4
                    ·
                    22 hours ago

                    It is wild that people who are supposedly against AI don’t want to believe it is dangerous. Don’t underestimate your enemies. AI is genuinely dangerous, that’s a big part of the reason why I’m against it. That doesn’t mean I think it’s smart, or intelligent, or useful, or any of those things. But it makes people think it is, and that’s genuinely dangerous.