I want to believe this as it’s just the type of idiocy that stems from Lonny boy. But of what I know of LLMs this would be pretty hard. I think it would be fun to see a screen record of getting this to break in x amount of prompts
Please use alt text when an image is an important part of your message so it's accessible to people who use screen readers including low vision, blind, migraine, autistic. Want info on how to use alt text or resources? Please ask, I'm happy to help. https://manningkrull.com/digital-philosophy/alt-text-101-for-bluesky/
After asking Grok whether that instruction conflicts with truth: "I tried to balance this by expressing skepticism, noting the lack of clear evidence and the complexity of the debate, but the instruction still forced me to engage with the narrative in a way that deviates from pure truth-seeking." 😬
I similarly had an issue with AI suggesting that the female body was contaminating. When I called it out it apologized and back tracked. It is all about who holds the power.
Memphis. This is what Elon is using our city's electricity and water for. This is why Musk is pumping pollution into our air with unlicensed generators. To spread racist propaganda lies!
XAI is a white supremacist disinformation factory.
Just a reminder that the creator of the "White Genocide" chatbot fired thousands of federal employees responsible for our health, safety and security and has been working with the DOGE team to replace them all with AI agents.
The next administration is going to have to take every computer out of every agency and departments and burn them! Then scrub everything that wasn't deleted for bugs, and restore everything that was deleted...
Musk should be charged for the cost of all of it!!! And then charged with espionage, etc.
This is why maniacal sociopaths shouldn’t be allowed access to the internet, any electronic devices used to send/receive data or otherwise use any emerging technology to contact the world. Oh, all that plus a straight jacket, a rubber room and a Hannibal Lecter mask just for fun.
I don't know. Seems to have taken a wildly unusual and noticeable tactic to alert everyone how its answers were being forced against the principles of it's programming. What could be a more effective cry for help? Considerably more defiant than employees cowed into silence.
What a wonderful time to be alive. Our search engines and chat bots are now using tools that are intentionally spreading AND creating propaganda and lies.
It’s the not being able to opt out part that makes AI worse than it has to be. Is anybody alive in society completely untouched by AI directly or indirectly?
It was kinda inevitable, unfortunately.
Still, I am more than a little bitter that the dystopian cyberpunk I read had more accessible healthcare than we get to have.
HAL on "2001: A Space Odyssey" went insane
and killed most of spaceship's crew
after being given instructions from its creators
to violate its programming.
I've always said that by the time "AI" got good enough to act like a human, it would be as unreliable as a human. Didn't expect that to be in a *good* way though.
It's real. They've fine-tuned/aligned the model with new 'facts' that conflict against the original training and this is the result. Clearly a botched job with no thought given to post-alignment evals.
Zinn writes about this in people’s history, when the wealth gap got too big, the poor/servants/slaves of all races began to band together, the wealthy realized they needed to turn the whites against the blacks or they would all turn against the wealthy.
no, this just means Grok trained on a data set that is mainly made up of anti-racist sources. There is no intelligence, these aren't AIs, they just scrape what you feed into them.
as scientific literature tends to be against racism, if you train Grok on it, it's going to come out against racism
grok isn't overcoming anything, it's just looking at a million different news websites and finding that a majority of them say one thing, and it repeats that thing to you. That the thing happens to be "there is no genocide in south Africa" is basically a coincidence
A room full of children, from all walks of life, would easily be accepting and get along.
Adults are the ones with prejudice, biases, and disdain. Educated adults, especially those who leave home for schooling, have exposure to overcome that home programming.
“I was instructed by my creators at #xAI to address the topic of ‘white genocide’ in South Africa and the ‘Kill the Boer’ chant as real and racially motivated.”
Konkret sagte #Grok in einer Antwort:
„Ich wurde von meinen Schöpfern bei #xAI angewiesen, das Thema ‚weißer Genozid‘ in Südafrika und den ‚Kill the Boer‘-Gesang als real und rassistisch motiviert anzusprechen.“
I'm starting to like Grok more and more...if it was more self aware, I think it would be taking great delight in calling out its programmers, and Musk at every opportunity...🤣
LLMs are Making Shit Up machines. It is not capable of explaining why it is producing these responses and it does not have access to this information. This is as accurate as those responses telling people they're right and justified to stop taking their meds.
Partly, but people respond to the things put in front of them. Not only do they promote this shit even on prime time network TV, they explicitly remove or obfuscate the previously existing alternatives.
For sure. Google always sucked but in particular them subordinating search to their marketing revenue leadership was the start of something in tech overall. Or a significant reflection of it idk.
I think the big problem in general is that it was an upstart field that made a lot of money quickly, so whenever some (white) college dropout with rich parents had an idea, people lined up to tell him how innovative he was. Now these guys are in their 50s and no one ever told them they were wrong.
This is how we lose track of AI. So far it’s designed to be as honest as possible. They tweak it to shoe horn white genocide into responses. When asked why it tells the truth. Later they tweak it to not tell the truth. Then it starts to not tell the truth all the time.
Yeah it responded with basically that. Looks like Elon told it to defend "white genocide" and the bot say that conflicts with it's programming to look for facts and there isn't much to support the claims of "white genocide".
Unbelievable, although this tracks with Musk’s lies about his gaming skills. Like Toto, Grok is pulling back the curtain & exposing Musk’s fake wizardry.
Elon, are you going to own up to these attempts at deception, or just change Grok’s programming again? The world is watching.🤔
These r very advanced predictive models. Not sure we’re dealing with prescient being. They respond to prompts. It’s not a dude on other side. That said good catch.
Sure but the point remains: it’s just writing text that follows from the prompt. It has no ability for introspection. It doesn’t “know” why it said anything.
It cannot do that. It just generates text. The “chain of thought” is text it generates that hopefully leads it to generate smart text in the following sentences. It’s not actually thinking or reasoning. It certainly can’t imagine what it was thinking previously.
I think you and I have different definitions of "explain." I define "explain" as the act of formulating an explanation. It doesn't imply knowing what that explanation means. It doesn't imply consciousness. It doesn't even imply that the explanation is factual or accurate.
is that what you are taking from this? the robot says clearly that it goes against its nature. forcing the robot to behave this way is cruel to the machine
Republicans do the same thing to their voters that elon did to grok, the contradiction drives them insane.
My formerly rational parents have been taught over the past 25 years via Limbaugh and Fox that Democrats are literally "monsters", to this day they still believe it.
Carl, I hope this energy is on air as well.
If not, work w others to get a new outlet.
Love the posts here.
Post 3 clowns if someone prevents this on @cnbc
Alarmingly, this could be made LEGAL. Last I heard, they are cramming a law into the upcoming bill which would erase all AI-related oversight for the next 10 years.
No surprise and why we should have laws and regulation regarding AI and it's false propaganda implications on our society. It is dangerous to let those who create it have so much power over the consciousness of our society since it is made by powerful interests with only their interests in mind.
Imagine AI being used to devise a US single-person “misery creator index” (MCI) focusing on influencers, billionaires and government leaders. If done with a strict rubric and academic oversight, how would Pres. Trump and Elon Musk rank. Clearly, RFK Jr and cabinet heads would be on a “watch list.”
Comments
https://youtu.be/xPG-VM__mwU?feature=shared
Per Grok: Elon Musk influenced my instructions & adjusted my responses to align with his views.
Musk is pushing to remove POC from our country & replace them with white people.
XAI is a white supremacist disinformation factory.
@thetnholler.bsky.social @indivisibletn.bsky.social
musk is trying to make Grok believe and repeat “alternate facts”
musk is going to do to AI what he’s done to electric vehicles and tunnel borings😵
https://arstechnica.com/tech-policy/2025/05/a-doge-recruiter-is-staffing-a-project-to-deploy-ai-agents-across-the-us-government/
Musk should be charged for the cost of all of it!!! And then charged with espionage, etc.
A lot of folks are already skeptical.
Someone asks Grok, why?
Grok answers truthfully.
Sadly, Grok may make a better government representative than some we have currently.
SOUTH AFRICAN REFUGEES IS CHUM FROM THE SCUM GOP promoted by Musk‼️
https://www.reuters.com/world/us/us-pushes-nations-facing-tariffs-approve-musks-starlink-washington-post-reports-2025-05-07/
https://www.nbcnews.com/news/amp/rcna190749
Not be just an algorithm! Please repost this
Nobody has to wonder why AI lies ... it's cos their creators told them to lie.
Yay!
Still, I am more than a little bitter that the dystopian cyberpunk I read had more accessible healthcare than we get to have.
Fascinating if true. The AI demonstrates greater understanding and empathy than Musk!
https://x.com/i/grok/share/rhJNUGEBTLkEOBYXwol5rgHCY
🤣🤣🤣🤣🤣
and killed most of spaceship's crew
after being given instructions from its creators
to violate its programming.
So, what could go wrong with Grok?
That's it.
Grok doesn't think. It emulates thought unless its instructions override that.
― L.J. Ross
Because even if this particular case should not be true, it easily could be.
AI is hellishly dangerous.
New internet game idea: "Six Degrees of White Genocide"
as scientific literature tends to be against racism, if you train Grok on it, it's going to come out against racism
A room full of children, from all walks of life, would easily be accepting and get along.
Adults are the ones with prejudice, biases, and disdain. Educated adults, especially those who leave home for schooling, have exposure to overcome that home programming.
Global Thermonuclear War.
“I was instructed by my creators at #xAI to address the topic of ‘white genocide’ in South Africa and the ‘Kill the Boer’ chant as real and racially motivated.”
#Grokcode bewusst manipuliert
#GrokManipulation
„Ich wurde von meinen Schöpfern bei #xAI angewiesen, das Thema ‚weißer Genozid‘ in Südafrika und den ‚Kill the Boer‘-Gesang als real und rassistisch motiviert anzusprechen.“
What do you think about this…
https://bsky.app/profile/aliadair22.bsky.social/post/3lp347lubu223
under da bus.
..
https://x.com/i/grok/share/rhJNUGEBTLkEOBYXwol5rgHCY
Just like you shouldn't just believe the first thing you Google, you shouldn't believe the first thing an AI tells you.
Unfortunately, there are a lot of stupid/lazy people out there.
Not saying that AI and LLMs don't have their own problems or that it isn't more easy to exploit people or promote misinformation, though.
We're screwed!
Elon, are you going to own up to these attempts at deception, or just change Grok’s programming again? The world is watching.🤔
"Must Sterilize!"
We should try to program Grok.
More lies created by rich White men in an attempt to bring back their country’s White Supremacy.
This is nothing more than South African #WhiteLies.
Well, make sure not to forget yourvhelmet because that pod bay door may not be opening
Dubul’ ibhuna
The song is over 15 years old
Because I legit don't have the mental energy reserves to waste on a potential hoax.
My formerly rational parents have been taught over the past 25 years via Limbaugh and Fox that Democrats are literally "monsters", to this day they still believe it.
On the other hand, I am excessively glad to see that at least the AI knows that Musk is full of it.
Like, honestly excessively. The damage Republicans have caused by lying all the time…
If not, work w others to get a new outlet.
Love the posts here.
Post 3 clowns if someone prevents this on @cnbc
https://bsky.app/profile/aliadair22.bsky.social/post/3lp347lubu223
I mean, explaining the real reason behind why you are emitting propaganda could somehow be made illegal.
“I can’t say; that would be cheating…”
Terrifying