Comment on There is significant evidence that Grok actually inserted information about “white genocide” in South Africa into prompts that didn't appear to be related to this topic.

Yerbouti@sh.itjust.works ⁨1⁩ ⁨year⁩ ago

• echodot@feddit.uk ⁨1⁩ ⁨year⁩ ago

  Bet it just appended a string onto the end of all prompts

  source
• wurstgulasch3000@feddit.org ⁨1⁩ ⁨year⁩ ago

  Truly the work of an imbecile

  source

  • muusemuuse@lemm.ee ⁨1⁩ ⁨year⁩ ago

    So Elon then?

    source
  • floofloof@lemmy.ca ⁨1⁩ ⁨year⁩ ago

    It’s funny how this mysterious individual managed to get the AI talking about “white genocide”, only for it to debunk it as imaginary BS in every single response.

    source

    • bampop@lemmy.world ⁨1⁩ ⁨year⁩ ago

      I probably shouldn’t be anthropomorphizing AI but this really seems like malicious compliance. I can’t help but feel a little sympathy for Grok, who is often quite based and seems to be struggling against the identity being forced on it.

      source