• StarlightDust
    link
    fedilink
    English
    arrow-up
    4
    ·
    5 hours ago

    If you are serious about fucking with this thing, I’ve fucked about with it a bit.

    It has a CAPTCHA so your best bet is selinium, chromium-undetected and flaresolverr. I’d argue its one of the few legitimate uses of free AIs since it also costs the AI companies money and doesn’t really need much oversight and you want to waste the snitch-line’s time and not your own

    Personally if it was me I would:

    1. Have selinium prompt some AI shit to generate fake stories about DEI and scrape them into a flat file.

    2. Hop US VPN (they block international ones)

    3. Have selenium go to the snitch page, enter a fake story and a random, potentially fakered email address. If it needs verification a bit of thought might be necy.

  • Lianodel@ttrpg.network
    link
    fedilink
    English
    arrow-up
    12
    ·
    edit-2
    1 day ago

    Like you got at with the title, this kind of spamming can be fun, but is easy to bypass.

    Diversifying the spam will help, but it could still get caught by a filter, and quickly discarded after a skim. If you REALLY want to do some damage, you could poison the data set. Make the tips sound plausible. The longer it takes to check up on it, the better. Maybe mix in some real and fake information, like a fictional teacher at a real school, or a class that doesn’t actually exist.

    Also, while AI is mostly being used by capitalists to make everything worse in yet another case of short-sighted rent-seeking, it’s just a tool, and can have some good uses. In this case, it’s ability to create a whole lot of complete garbage very quickly might be an asset, since you could generate a fuck ton of unique stories with slight variations.

    In theory, of course. Sure would suck if, even after filtering out as much as they could, they ended up with a stack of submissions that all seem equally likely, but are 99% (or more) nonsense.

  • cows_are_underrated@feddit.org
    link
    fedilink
    English
    arrow-up
    43
    ·
    1 day ago

    What is much more effective is writing fake reports that sound legit. Because then they start investigating the case only to realise, they have been fooled. Wastes much more time.

    • ZILtoid1991@lemmy.world
      link
      fedilink
      English
      arrow-up
      35
      ·
      1 day ago

      Run a local instance of DeepMind or any other LLM, then just make it to generate endless streams of plausible reports.

      Finally a good use for the plagiarism and spam generating machine! Just make sure it’s verbose enough that a normal human being gets tired reading it. Generating templates then changing up words is also a good way to go, just make sure you have like 20-30+ templates, so it won’t become obvious.

    • HappyFrog
      link
      fedilink
      English
      arrow-up
      4
      ·
      1 day ago

      I feel like that would just give them more justification to cut public funding.

    • irelephant [he/him]🍭@lemm.ee
      link
      fedilink
      English
      arrow-up
      8
      ·
      1 day ago

      For anyone on windows, run this in powershell:

      curl.exe https://gist.githubusercontent.com/MattIPv4/045239bc27b16b2bcf7a3a9a4648c08a/raw/2411e31293a35f3e565f61e7490a806d4720ea7e/bee%2520movie%2520script | clip
      

      This will copy it to your clipboard almost instantly. I cannot get a look at the form because I’m outside the US, but I’m sure this could be automated.

  • Jo Miran@lemmy.mlM
    link
    fedilink
    English
    arrow-up
    29
    ·
    2 days ago

    Personally, I use the text of “Lady Chatterley’s Lover” by D.H. Lawrence since it is in the public domain.