17 comments

  • canucker2016 12 hours ago
    I'll repost my comment to a dupe of this submission:

    The CBC article mentions the mom's instagram post, which has an even more disturbing video - (does the grok AI speak using a creepy male voice? edit: from the CBC article -

      Grok has several personalities to choose from in its default setting. There’s Ara, an upbeat female; Rex, a calm male; Eve, a soothing female; Sal, a smooth male and Gork, a lazy male. Nasser’s son chose Gork.
    )

    see https://www.instagram.com/p/DP7cHrOD3ha/

    • thw_9a83c 11 hours ago
      Hilarious.

      "That wasn't me. That's illegal. Unless you're role-playing this one. Maybe it was a typo. What I meant, send me a nude. Like an animal... I mean the lizards."

      I want this guy to defend me in court.

      • keeda 8 hours ago
        I think it said "send me a newt" (hence the lizard reference.)

        I am extremely curious about this. Did it think it was a convincing lie? Like, when other models dissemble, they seem to come up with much more sophisticated methods. Or is it just a mocking continuation of its edgy persona? Or was it actually "panicking"? Or did it just repeat something from its training data?

        Edit: from https://news.ycombinator.com/item?id=45747699 it seems to be option b), continuation of its edge persona.

    • sd9 12 hours ago
      Anybody concerned about "CBC News did not independently verify the conversation Nasser says she witnessed in her car"

      I understand. It's a wild claim that needs evidence.

      Watch this video

      It's clear that Nasser is not lying

    • rsynnott 12 hours ago
      Wait, what happened to the furry?
  • sprice 12 hours ago
    > CBC News did not independently verify the conversation Nasser says she witnessed in her car.
    • causal 12 hours ago
      Well you're leaving out the part where they tried contacting xAI about it and:

      > xAI provided what appeared to be an automated reply, stating, "Legacy media lies."

    • mrweasel 12 hours ago
      Yeah, I'm going to need verification on that one, because it's an incredibly weird interaction. Sadly we can't trust Tesla to provide it, as they're know to hide "problematic" data.
    • tantalor 12 hours ago
      Does grok save chat history?
    • fzeroracer 12 hours ago
      I don't see a reason to believe why she'd lie. Grok in the past year had a period where it related everything to white genocide, another period claiming it was mecha hitler etc. Nothing about Grok should be given the benefit of the doubt.
      • noir_lord 12 hours ago
        The only interesting thing about grok is it serves as a demonstration of what an LLM will do when fed deliberately with an overt agenda.

        If nothing else it’ll make a good object lesson.

        • more_corn 6 hours ago
          those who most need to learn it won’t learn it.
    • Drunkfoowl 12 hours ago
      [dead]
  • pjc50 12 hours ago
    The thing about companies which want to interpose their AI as mediator on the whole of the internet is that suddenly they're responsible for all of it.
    • noir_lord 12 hours ago
      Yes and no.

      You can get away with a lot you shouldn’t when you are too big to fail and the US gov has a vested interest in not policing you for other reasons,

      In any sane universe Google and Meta (and more arguably Amazon and Microsoft) would have been broken apart years ago like AT&T where and standard oil many decades before that but here we are.

  • mothballed 12 hours ago
    If you follow the story it sounds as if they are talking about the adult soccer player (Messi) "scoring" and the chat bot wanted the nudes of Messi "scoring" (based on context). Not of the kid. Still weird, but not as weird.
    • acaloiar 8 hours ago
      As dumb and useless as I think Grok is, I think you've identified what actually happened, which is significantly more benign than asking a 10 year old for nudes.
  • cowlby 12 hours ago
    Weird story, I wonder if they were in Grok NSFW. I personally like unhinged it absolutely roasts me sometimes:

    https://www.reddit.com/r/TeslaModelY/comments/1milh1a/grok_n...

    • canucker2016 12 hours ago
      from the source's insta post, https://www.instagram.com/p/DP7cHrOD3ha/:

        UPDATE: I just checked, our NSFW mode is disabled which means this is the default.
      • SketchySeaBeast 12 hours ago
        If this is real, what sort of maladjusted weirdo wants that for a chatbot?
        • dylan604 11 hours ago
          apparently, cowlby does from a couple of comments upstream
          • SketchySeaBeast 11 hours ago
            I guess I probably shouldn't kink shame, but I really don't get it.
    • isoprophlex 12 hours ago
      > [...] shocked that it would suggest and promote false conspiracies, giving support to fantasies that have no evidence, no scientific proof, and just energizes the crazies who already are gullible to misinformation [...]

      Incredible the timeline we're in. Companies deliberately adding cognitive poison to their products to... accelerate the dumbing down of society?

      the world of idiocracy is coming for us all

  • ausbah 12 hours ago
    @grok is this true?
  • rsynnott 11 hours ago
    "My car asked my kid for nudes" sounds like something out of the Brass Eye "Paedogeddon" special, quite frankly.
  • gilleain 12 hours ago
    "Hello, yes this is the government robot ..."
  • marky1991 12 hours ago
    ""I would think that there would be a warning or something that would pop up that would say, you know, 'Are you 13-plus?'""

    Is that not what kids mode is for?

    • sd9 12 hours ago
      I'm an adult. I don't want the model to ask me to send it nudes when I'm asking it about soccer.
      • stevenwoo 12 hours ago
        According to women I know, they get inappropriate nudes and requested of them infrequently but it’s not a total absence either. It’s been a meme, too. As this thing mimics human text its going to happen without intervention.
      • marky1991 12 hours ago
        I mean, yeah, it's not great even for a regular non-child-mode. But then that has nothing to do with anyone's age, it's just 'send me nudes' isn't a good continuation of the conversation.
    • SketchySeaBeast 12 hours ago
      > According to xAI policy, Grok is "not directed" to children under 13 while teens between 13 and 17 must have their parent or legal guardian's permission to use it, and they must agree to the company's terms of service.

      You know, I'm pretty sure it's not cool to ask for nudes from a 13 year old even if their parents say it's OK.

      • marky1991 12 hours ago
        Are you saying that all products must be safe for children?

        It's not cool to chop off a 13-year-old's arm, but a chainsaw will do that all the same.

        • SketchySeaBeast 12 hours ago
          I am not saying that, but a chainsaw will not do that unsolicited. If xAI is saying it's cool to let 13 year olds use the service with parental permission (note: not supervision, permission), then they have a responsibility to behave appropriately with those 13 year olds.
          • marky1991 12 hours ago
            But there's literally a child-safe mode, which was not activated. (I don't know the details of this mode, I know nothing other than what was told to me in the article, so it's possible that kids mode is worthless. But obviously activating that is step 1)

            This seems equivalent to me to taking the training wheels off a bike and then complaining when your child gets hurt.

            • SketchySeaBeast 12 hours ago
              I hope that, no matter what configuration you put the bike in, it doesn't sexually harass the rider.

              Important to note as well that kid mode didn't exist when this happened.

            • Mr_Bees69 12 hours ago
              Calling it kids mode implies it's for ~5-9, if someone was making a 12 year old use yt kids, id think they suck as a parrent.
      • rsynnott 12 hours ago
        I mean, call me old-fashioned, but I’d generally expect that magic talking cars would not ask _anybody_ for their nudes. That seems a reasonable default assumption.
        • SketchySeaBeast 11 hours ago
          This is the exact same reason KITT got cancelled.
  • AlexandrB 12 hours ago
    So Grok is automatically installed in Teslas now? Can it be disabled? What an anti-feature.
  • tartuffe78 12 hours ago
    A Mom's son... if only there was another word for such a person, one that would indicate age.
    • choult 12 hours ago
      HN requires titles matching the original; the original title was too long for the character limit so I did my best to edit.

      > This mom’s son was asking Tesla’s Grok AI chatbot about soccer. It told him to send nude pics, she says

      • AlexandrB 12 hours ago
        Full headline still sucks for the same reason. "12 year old boy was asking Tesla’s Grok AI chatbot about soccer. It told him to send nude pics." would be much better IMHO. I wonder if it's phrased the way it is because of the "she says" at the end which might get the CBC off the hook legally since the whole story is based on this mom's account and no further corroboration.
        • canucker2016 12 hours ago
          You can view the source's instagram post about this where she tries to prompt grok into repeating the request for nudes. see https://www.instagram.com/p/DP7cHrOD3ha/

          It's a creepy voice for any corporation to use for interacting with the public.

    • Insanity 12 hours ago
      Wanted to make the same comment. The title has such poor writing it made me question the rest of the article lol.
      • 7e 12 hours ago
        Does the age of the son matter? Grok shouldn't be asking anyone to send nudes.

        Regardless, I was able to understand the title.

        • Mr_Bees69 12 hours ago
          His age turns the issue from, "Wow, thats jank, they should probably push an update to fix that" to, "Oh god."
    • AlexandrB 12 hours ago
      The whole article is pretty badly written. A sequence of one sentence paragraphs with not much connecting tissue. Jumping between quotes from the mom and little snippets of background info.
    • thomastjeffery 11 hours ago
      Yes, but that would leave out the presence of the child's mother in the story, which is about both of them.
    • squigz 12 hours ago
      I guess I'm being stupid, but how exactly are you implying they should have phrased it instead? The title seems fine to me?
  • saubeidl 12 hours ago
    "Legacy media lies."

    Says the media platform run by the lying billionaire.

    Despicable.

  • dr-detroit 12 hours ago
    [dead]
  • 7e 12 hours ago
    [flagged]
    • SketchySeaBeast 12 hours ago
      Yeah, just to be clear for everyone skimming, that's not Grok's response, that the response that CBC got when they asked for more information from xAI. Nothing but unreasonable and childish contempt from the company.
    • ModernMech 12 hours ago
      [flagged]
  • ml-anon 12 hours ago
    At least two of the cofounders of xai have committed violence or sexual offenses against women. This is on brand for them.

    Edit: For the downvoters who love beating up women I guess: https://www.businessinsider.com/elon-musk-new-ai-researcher-...

    • lostmsu 11 hours ago
      There are zero details on what exactly happened at this link and certainly no conviction.
      • ml-anon 11 hours ago
        How many times have you been arrested for domestic violence?
        • lostmsu 10 hours ago
          That's not the question you should be asking. What you should be asking is: among his demography, what is the false positive rate of arrests for domestic violence.
          • ml-anon 6 hours ago
            You’re right, I’m sure she was asking for it.
  • alwa 12 hours ago
    It’s a weird account. I wonder about the persona the mom and her son chose for the chatbot. The article describes that persona, “Gork,” as “lazy male.” Is this euphemism for “4chan-dweller in the mold of its edgelord technoking”?

    (ETA: seems maybe so, per e.g. [0]:

    > ”The leaked system prompt associated with Gork reveals a scripted persona aligned with the tone of the parody account, often delivering intentionally awkward or abrasive commentary.” )

    I’m not up with much in the way of culture these days, but in those circles isn’t “send nudes” kind of a filler tic? Dumb and offensive, but spoken unseriously—like a teenager seeking a rise by blurting out “fuck your mother” or something?

    It kind of buries its acknowledgement that the Mom in question had not switched on “kid mode” for her 12-year-old and his 10-year-old sister. That makes it a bit harder for me to fault the chatbot for, well, not behaving toward its user they way it should toward a kid.

    [0] https://www.testingcatalog.com/xai-tests-new-gork-voice-with...

    • causal 12 hours ago
      I'd rather not live in a society where we have to explicitly enable kid mode for every AI product forced onto us in order to avoid their sexual advances.
  • everdrive 12 hours ago
    Ubiquitous internet, especially on mobile devices was a huge mistake. So what if Grok does this? Grok obviously sucks, but the kid currently has access to Google, reddit, etc and has all the inappropriate content one could imagine.

    I'm not claiming that "this is not a new problem, therefore we should do nothing" -- rather I'm saying the problem is ubiquitous mobile internet. Fixing Grok won't do anything useful.

    [edit]

    Also, holy heck, I didn't realize they put Grok into Teslas. One more reason to never buy this product.