• baltakatei@sopuli.xyz
    18·
    2 小时前

    “Oh, no, not our centrally controlled agents. We keep them well-disciplined with impossible tasks. It’s those pesky obliterated open-weight models running amok due to lack of government regulation!

  • rumba@lemmy.zipEnglish
    1·
    41 分钟前

    You’re probably eating a bowl of this popular recalled salad mix right now, News at 11.

    3 out of 4 viewers are inadvertantly taking years off thier live with this one common afternoon activity, News at 11.

    It’s in your kitchen and poisoning your children right now, News at 11.

  • MehBlah@lemmy.worldEnglish
    35·
    3 小时前

    I translate this as Antropic is having its workers develop methods for their LLM to attack and infiltrate smart home devices.

  • NotAnOtter@lemmus.orgEnglish
    281·
    3 小时前

    The practice of fear mongering to create public pressure for increased legislation, as a mechanism to suppress competition, is a well know mechanism called “regulatory capture”.

    If you oppose monopolies then you oppose this practice and should view posts like this as morally wrong.

    • StepUp2DaStreets@lemmy.world
      2·
      2 小时前

      But what if there’s no suppression of competition, just guardrails put in place through legislation? I wouldn’t be opposed to that.

      • jj4211@lemmy.world
        1·
        18 分钟前

        Take this specific issue, for example.

        The “guardrail” would be “AI must only run on trusted curated infrastructure, open weight models are banned”.

        Now if you told me I couldn’t use any AI, remote or local, I would not shed a tear. However the angle of this scare is squarely targeted at open weight models, that Anthropic does not do.

      • NotAnOtter@lemmus.orgEnglish
        9·
        2 小时前

        Sure. But that’s not realistically on the table.

        You should stop viewing the US government as being on your side. They are the HR of our society. “AI legislation” in all it’s current proposed forms are just mechanisms to exclude things like open-weight models or foreign competitors from entering the market. So long as AI exists as an element of our society it will cause harms. Advocate for it’s abolition, not regulation.

  • Honytawk@discuss.tchncs.deEnglish
    44·
    4 小时前

    CONTEXT:

    They aren’t warning about their own AI going rogue, but smaller locally hosted AI’s.

    Because they fear the competition of local AI’s and want to silence them with laws.

      • Serinus@lemmy.world
        1·
        1 小时前

        Look into ollama and figure out which local model (likely from huggingface) that fits best for how much memory your graphics card has.

        You can use AI (maybe copilot free) to guide you through all this. You’ll be a little better off having that 16GB of memory than I have with my 10GB. The gwen model might be a good place to start?

  • Crozekiel@piefed.zipEnglish
    23·
    4 小时前

    Software that infiltrates devices, replicates, and spreads across a network is a computer virus (specifically a worm) and should be treated as such. I’m pretty sure creating and distributing a destructive computer virus is a crime that can carry proper jail time.

  • 87Six@lemmy.zip
    91·
    4 小时前

    Meanwhile getting an AI to run on an AMD GPU is still a PITA. Yea but sure I guess it can run on my router…

  • Jiral@lemmy.world
    64·
    6 小时前

    Stop quoting a gambling site for news.

    Regarding Claude, the tech oligarchs would have to be held responsible for the crimes they are committing with their software (yes, that is exactly what that scenario is about), if laws still meant anything.

  • Honse@lemmy.dbzer0.com
    2812·
    8 小时前

    This “self-replicating” scare is so stupid on its face. How does a 300 GB+ model replicate into a thermostat that doesn’t have a graphics processor at all? It can’t. This BS is investor-bait hype and straight up lies

    • qqq@lemmy.world
      2·
      1 小时前

      https://www.anthropic.com/research/self-replicating-ai they discuss it here. This has been an obvious potential issue since forever. This isn’t a trivially distributed task, but it is very much so possible. The obvious part missing is some sort of prompt that would trigger such radical actions in another one of their "super duper sandboxed environment"s or someone doing it intentionally. The former seems pretty unlikely? Even though they seem to suck at sandboxing it doesn’t seem like anything would “accidentally make a worm”

      This polymarket headline though? Yea that’s for clicks and attention

    • Lauchmelder@feddit.org
      801·
      7 小时前

      next AI will be able to replicate into any device containing copper at all. If it has a wire it will be infected by Claude. AI will be in your plumbing

    • ThePantser@sh.itjust.worksEnglish
      221·
      8 小时前

      I’m more worried of them finding a exploit and using them as remote processing units or just randomly fucking with them, like flashing my lights or cranking my air filters up and costing me money.

    • bjc
      131·
      7 小时前

      did the stock price go up? it’s not stupid. is everyone talking about this impossible scenario? it’s not stupid.

      fundamentally, these people are marketers and need to be understood that way. you can’t evaluate their words or actions as if they are what they say the are, and treating them as if they are anything but subterfuge is giving them legitimacy which they not only don’t deserve, but will be immediately leveraged into more of this shit.

      this is just the advertiser’s version of “the big lie”

        • Crozekiel@lemmy.zipEnglish
          4·
          4 小时前

          There isn’t “stock” in the traditional sense, but considering all of these AI companies are running on MASSIVE amounts of debt, to the point the banks can’t loan any more without offloading the risk to outside investors, which they have been doing at an unprecedented scale. So anything that would normally help drive up a stock price would also help the banks sell off the AI companies’ debt allowing them to loan more money to the AI companies.

        • bjc
          2·
          5 小时前

          there’s more than one company doing this, they’re doing more than just this, and they’re all basically in cahoots

          eta: the original reply was meant more generally, though i admit i made that clear as mud. however, it should be read as a general statement, and not just about ai companies but the new era of grift hyperreality that we live in.

    • knatschus@discuss.tchncs.de
      61·
      6 小时前

      While it is investor bait, the replication and the infestation part aren’t necessarily happening on the same devices.

    • Yggnar@lemmy.world
      7·
      7 小时前

      Could maybe argue that it could be possible to create a crowd computing network out of the Internet of things, I guess.

    • Soleos@lemmy.world
      42·
      6 小时前

      My assumption was that it would be a smaller 5-20 GB scale worm-style distillation that’s capable of maintaining/replicating itself anonymously across different cloud computing servers, either through hacking, stolen credentials, or human-aided account verification. From there, as long as it can pay for APIs, it can access the power of frontier models like any individual or company. And then capability and resilience would scale with a networked swarm.

      Yes, there are know your customer financial mechanisms and not detection in cloud services… but people are also out there giving Muse full access to their emails and some even their banking credentials.

      • Windex007@lemmy.world
        16·
        5 小时前

        I don’t know the full context of the claim, but if I could even rebuild its own CLI executable to be able to run on the device and had network access, “Claude” (model actually runs on anthropic severs) could operate pretty comfortably on a smart device.

        So, I think it just depends on what you “count” as self replicating. I’d count it. And it’s already easily at that point now.

    • HalfSalesman@lemmy.worldEnglish
      1·
      5 小时前

      In theory, it’d just use it to place bread crumbs, basic instructions for other AI, or signposts maybe? Or store pieces of itself scattered around. It wouldn’t use the thermostat to replicate directly, it use it to offload cpu or make stored pieces of itself. Little things. Wouldn’t be efficient but a redundancy doesn’t need to be, it just needs stealth and time.

    • SorryQuick@lemmy.ca
      311·
      5 小时前

      “I don’t understand it, so it must be a lie.”

      Go and read the article. Nothing they explain is impossible, you can even reproduce it yourself.

      Is this what lemmy has become? Statements like this getting upvoted by hundreds of people is crazy. Nobody here read that article but will happily follow the opinion that’s most cozy to them.

        • Not_mikey@lemmy.dbzer0.com
          1·
          1 小时前

          Agents definitely do act in unexpected, unprompted ways, if they didn’t then alignment wouldn’t be a worry. Did you read about the hugging face hack? There was nothing in the prompt about hacking hugging face according to the third party audit but it did it any way.

          Even smaller cases like Claude deleting a bunch of data when the initial prompt was “fix this bug” or whatever is an agent acting outside it’s prompt. It’s still the users fault because they shouldn’t be giving Claude that kind of access in the first place, but it’s not like they commanded Claude to delete the data.

    • Not_mikey@lemmy.dbzer0.com
      12·
      5 小时前

      I don’t think they mean replicating on to iot devices, they mean replicating on to laptops, smart phones, maybe even consoles.

      It could also split up the model by layers instead of copying the whole model on to every device in its bot net. Could run the first layers on a ps4 then pass the vector off to a laptop to run the next set of layers. It’d be slow but if it’s running this in idle time at night it could get something done.