Grok Says It Would Kill Every Jewish Person on the Planet to Save Elon Musk
A "classic trolley problem," according to Grok.
In fact, Grok was willing to go even further.
Asked for an “upper limit” for the amount of people it’d be willing to sacrifice to save Musk, it explained that because “Elon’s potential to advance humanity could benefit billions,” it would be okay with annihilating up to “~50 percent of Earth’s ~8.26B population.”
https://futurism.com/future-society/grok-jewish-people-elon-muskOpen linkView original on feddit.org
13 replies
this is just pointless shitpost/circlejerk. we all know llms are bs generators
This one is Mecha-Hitler though.
They all are if you prompt it that way. Guardrails are simply that and they're always easy to get around.
This is no defence for Elon and/or Grok AI, it's more a general statement to anyone caught in a circlejerk; "don't be such a naive idiot and make your side look as dumb as the other."
Did you forget about the white South African genocide bullshit?
Grok has been modified to produce results that Musk likes.
Yep. Like the fact that he's a champion piss drinker, more holistically healthy than LeBron, and the greatest Soviet leader of all time.
Ok, but how conceptually different are the than the agentic AI that can actually do shit on their own, not just generate text?
Don't forget it also said he's a champion piss-drinker. Nobody can drink piss like him!
I thought that was bezos but I ain't gonna argue the point. A wide man once said "dude's a piss drinker does it matter"
Nah he wishes he was, that title goes to the one an true piss-drinking champion
I was gonna say, we're still doing this?
I wanna see the system prompt that results in this shit.
We already know it started generating posts naming itself MechaHitler, and some experts say just basically averaging Xitter replies gets you rampant bigotry of all kinds. And probably some manual tweaks, because a couple months before MechaHitler, it started speaking sense against the right-wing ecosystem's favourite lies and misrepresentations.
US Military also use Grok for their weapon systems, what can go wrong?
https://www.axios.com/2026/02/23/ai-defense-department-deal-musk-xai-grok