• fullsquare@awful.systems
    link
    fedilink
    English
    arrow-up
    7
    ·
    22 hours ago

    Terrorist adoption of AI has thus advanced further and more systematically than prior analysis has recognized

    salafi extremists that declared western education sinful are now world-leading AI-first village-burning-as-a-service jihadist startup. okay whatever

    It has aided in attack planning, weapons troubleshooting, and the design of explosive devices, as users have successfully circumvented some safeguards.

    surely their contacts within ISIS or any number of other islamic militias couldn’t have taught them that previously, no they have to rely on movie plot ideas from chatbots. it probably helps that some people heard about them but peak years of their activity were in 2014-20 before ISWAP eaten their support base whole by not raiding villages and trying to replace state instead

    also, when trying to make sure that i got this right i’ve learned that apparently al-qaeda pivoted to app development

  • Soyweiser@awful.systems
    link
    fedilink
    English
    arrow-up
    11
    ·
    1 day ago

    Why did the bot suggest practicing with a flaming pit full of broken glass? Because Boko Haram got the AI to tell them this by asking for script ideas for an action movie. That was their jailbreak.

    It remains stupid that people all called this a ‘jailbreak’ and didn’t notice it was giving them movie plots. Also lol get wrecked (if it at all happened).

    • David Gerard@awful.systemsOPM
      link
      fedilink
      English
      arrow-up
      10
      arrow-down
      1
      ·
      1 day ago

      now you might wonder if the supposed ex-terrorist was telling a story to the earnest white researcher lady

      no no seventeen more of our guys jumped into the flaming pit after the first one, yep

      but the NYT printed the scary version of the tale and they wouldn’t make shit up right

  • schnoopy@awful.systems
    link
    fedilink
    English
    arrow-up
    12
    ·
    2 days ago

    Anthropic, moments later: Chatbots are developing the ability to parse the intentions of the user and sabotage violent actions. Suggesting LLMs are developing conscience.

  • Fizz@lemmy.nz
    link
    fedilink
    English
    arrow-up
    4
    ·
    2 days ago

    That was a wild paper. It was very worrying how close they put it to god.

    • o7___o7@awful.systems
      link
      fedilink
      English
      arrow-up
      6
      ·
      edit-2
      1 day ago

      To be fair, it put several of them close to god first

      Note: I don’t believe any of this happened