Elon Musk's artificial intelligence (AI) chatbot appears to have published messages without users' knowledge.
  • AmidFuror@fedia.io
    38·
    3 months ago

    Ah, yes. Ask Grok to come up with a secure password that only you and everyone else know.

    • LifeInMultipleChoice@lemmy.worldEnglish
      19·
      3 months ago

      I do wonder how many people have copied config files to diagnose issues into Grok/other AIs not remembering they had IP addresses, certificate security information, or maybe even server credentials

    • felsiq@piefed.zipEnglish
      111·
      3 months ago

      I watched someone vibe code a curl request with grok and it spat out an api key in the configuration completely unprompted. It wasn’t an active one unfortunately (that would’ve been so funny) but I could totally see this happening

    • gdog05@lemmy.worldEnglish
      7·
      3 months ago

      “Please enter your password and the name of a random stranger for password authentication.”

  • panda_abyss@lemmy.caEnglish
    282·
    3 months ago

    Unique links are created when Grok users press a button to share a transcript of their conversation - but as well as sharing the chat with the intended recipient, the button also appears to have made the chats searchable online.

    This is t he r second article about chatbots “leaking” ended users share their chats with search indexing enabled.

    That isn’t leaking. That’s just shitty journalism.

    I hate grok.

    • KairuByte@lemmy.dbzer0.comEnglish
      3·
      3 months ago

      If the user clicked a “generate me a share link” button, and the button also, without letting the user know prior to the button press, enables search indexing, that is indeed a leak.

  • 60d@lemmy.caEnglish
    243·
    3 months ago

    What gets me laughing is when people use any technology and expect some level of privacy, especially “free” services.

    Like, ‘Oh noes! This new thing I shared all my deepest secrets with is now posting them online without my consent!’

    Milady, you consented by clicking okay. There’s no takesy-backsies.

  • Petter1English
    15·
    3 months ago

    👀 where is the database with all those leaked chatbot conversations?!

    Guess, it is pretty valuable