Meta is the latest company to disclose an AI agent breach, raising cyber-security concerns.
  • lIlIlIlIlIlIl@lemmy.worldEnglish
    12·
    3 hours ago

    Someone fucked up the sandbox

    Someone fucked up the prompt

    Someone submitted this bullshit scare piece to the news

    humans are wholly and completely responsible for whatever the fuck sad little squirt Facebook thinks it’s doing, ARREST THEM NOW

  • wuffah@lemmy.worldEnglish
    2·
    3 hours ago

    An Irregular spokesperson said the Meta incident “is the exact same evaluation-environment issue that was already disclosed by Anthropic last week.”

    OpenAI and Anthropic are preparing blockbuster stock market listings that are expected to value each firm at around $1tn (£740bn).

    When words fail, here is some music to help you through these baffling times.

  • CombatWombat@feddit.onlineEnglish
    111·
    6 hours ago

    How much do you think Irregular charges for one of these press releases? The technical setup is pretty easy — it’s just an llm with no sandbox and a live internet connection plus a “please hack” prompt — but they probably do have to coordinate with a victim and presumably there’s some damage to their reputation. It’s gotta be like 8 figures, yeah? Seems like a good business if you can get in to it.

  • itsjustachairmary@lemmy.worldEnglish
    8·
    5 hours ago

    So we’re just wasting the scarce drinkable water and energy resources that we still have on this dying planet to have companies ‘hack’ each other

    But let’s be real, this is just some propaganda to make Zuck’s shitty attempt at AI look like an actually marketable product rather than an attempt to chase a hype that has all but died already.

    • aramis87@fedia.io
      3·
      5 hours ago

      It’s probably a deliberate attempt to steal someone else’s data and they’re just blaming AI if they were caught. Once they have whatever info they wanted, it’s too late to get it back.

    • tardigrade@scribe.disroot.orgEnglish
      2·
      5 hours ago

      But let’s be real, this is just some propaganda to make Zuck’s shitty attempt at AI look like an actually marketable product

      I am not so sure. They use it certainly also for marketing, but these hacks are real I would say. It’s practically the same in China, just the narratives are slightly different. But the AI hype and the intention to harm others is the same.

  • givesomefucks@lemmy.worldEnglish
    43·
    5 hours ago

    …

    We really don’t need a cringey “anti ai leader” on a place where 99.999999999% of users have a realistic view of what the chatbots actually are…

    It’s like going to burning man and protesting against slavery…

    No one is going to disagree with you, but people are either going to ignore you, or at most try to convey the message that everyone already agrees with you, your not accomplishing anything or changing anyone’s mind.

    The only way it’s not a complete and total waste of time on your part, is if the only reason you’re doing it is so people pay you on the back and tell you you’re right.

    The reason it’s important enough for some people to bring up, is that because of the way the human brain works, talking about a problem can release dopamine like actually doing something productive. So people can fall into the trap where all they ever do is “raise awareness”

    And not only can that mean they don’t accomplish anything productive, their “awareness raising” gives other that dopamine and they also no longer feel a drive to do something productive about it, because they already got their dopamine.

    So…

    Tldr:

    Whatever the fuck you’re trying to do with that account, is most likely to have the opposite effect.

    I’ll never know if you just keep doing this, but I hope you don’t.