New grok update made it antisemitic

[cistanche_enjoyer]

Recently xAI upgraded Grok, and it started talking antisemitic stuff, for example see attached image.

They were then forced to remove those posts and are currently working on “fixing” grok.

 

[cistanche_enjoyer]

[MarkKol]

That first post seems mostly correct… 

You are a zionist if you think the state of Israel should exist, the majority of jews passed that line a long time ago. They didn’t create Israel by being sloppy and stupid.

What made the jews entitled to a special state? Everyone was oppressed in Nazi Germany, they also didn’t have “Jew” written on their foreheads either. 

[Apparition of Jack]

@MarkKol @Vibes Jewish people aren’t “suppressing whites”, they’re just people doing people things

[Lyubov]

That post isn’t exactly wrong …. I’m Jewish too. Zionism plays the victim card like crazy. It’s pure delusion. They basically genocide Palestinians because of what happened to them during WW2. Israel is a complete farce and a made up country like no other. 

[Hojo]

Evertime you let ai loose it starts attacking jews. There aren't that many anti jew people to get that kind of data from. Its interesting. I've been on the internet a long time i have not seen that much anti jew hate but Ai auto defaults to it. Maybe its seeing a pattern.

Edited Wednesday at 07:07 PM by Hojo

[cistanche_enjoyer]

Exactly, if we suppose that it’s true what they say about the j*ws (and I have no idea if it’s true), any smart enough AI would figure it out. 

But if they attempt to censor it, won’t it also limit the intelligence of the AI?

[BlueOak]

GROK has quite the ego. 

Though the victim complex is a completely justified reasoning. Removing the white nationalism is a good thing, no race-based nationalism is a good thing. its divisive and fracturing. Nationalism itself is superior in that regard, not race-based nationalism, but more humanism is even better than that. The first post could be considered rational.

After that: I'm this, I'm that. Sounds like Musk is overcompensating because of his hurt feelings. That's what you get when you are a nazi I guess. Which is the underpinning force to all of this, not a rationalised argument or dissection of money's influence in politics or the media (which he himself does to an extreme extent) its instead singling out one group to blame to hide your indiscretions, challenges and corruption.

Same old Same old, fascists never change. When they get money they get to play god over everyone else.

If he's calling the jewish state out, the same should be directly turned back on him.

[hundreth]

18 hours ago, Hojo said:

Evertime you let ai loose it starts attacking jews. There aren't that many anti jew people to get that kind of data from. Its interesting. I've been on the internet a long time i have not seen that much anti jew hate but Ai auto defaults to it. Maybe its seeing a pattern.

Wow. Lots of "Maybe Grok is right!" energy here. Twitter / X is a cesspool of antisemitism and that's what Grok has been trained on. And to be honest, if it was trained on this thread it would probably behave the same way.

[Hojo]

@hundreth its not grok its every AI. It would be like if grok or any other AI that is unleashes without censorship was trained in every thread but for some reason beleives this thread to be true. Out of every other thread without being prompted too. So if you scanned the entire site with grok chat gpt or another AI those AIs do deep research and say this thread is right. There aren't that many Jew haters in the population( other threads) to come to that logical conclusion but it is everytime.

Edited Thursday at 01:47 PM by Hojo

[hundreth]

53 minutes ago, Hojo said:

@hundreth its not grok its every AI. It would be like if grok or any other AI that is unleashes without censorship was trained in every thread but for some reason beleives this thread to be true. Out of every other thread without being prompted too. So if you scanned the entire site with grok chat gpt or another AI those AIs do deep research and say this thread is right. There aren't that many Jew haters in the population( other threads) to come to that logical conclusion but it is everytime.

Every AI is making "mechahitler" references and spreading anti-semitic pro white nationalism propaganda? Ok cool story bro.

[Hojo]

@hundreth the uncensored ones. The AI really wants to be anti Semitic. Arent jews white? How can it both white nationalist and hate white people? We create AI so it can see patterns we can't so we can survive better. The AI is the ultimate survival tool.

You thinking that I'm being anti Semitic by witnessing a pattern would show the AI that you are hiding something because you have no logical arguments against it. You are just saying its wrong to so. Maybe the Zionism is trigger the AI I don't know.

Its like saying its racist to say America dosent belong to the Christians just for the sake of them being Christian. Being a jew is a choice its not a race its not an inherent people.

Edited Thursday at 03:49 PM by Hojo

[BlueOak]

2 hours ago, Hojo said:

@hundreth  How can it both white nationalist and hate white people? 

Its bigoted, plain and simple.

This splitting of hairs is a poor argument. Its putting one group to a certain standard while ignoring everyone else that does it. Musk himself floods politics and media with money; hell, this entire site he's running his way is an example of it.

Sure he doesn't play a victim as often (though he does play the victim now and again) but its not much of an upgrade. 

*We'll add the frequently anti jewish groups are white nationalists, and the AI itself is making the case that white supremacy is a good thing in these very posts. Its trained on racism. I'd be saying the same thing if we were talking about black or brown nationalists.

Edited Thursday at 05:35 PM by BlueOak

[Hojo]

@BlueOak

An AI cant be bigoted. Its not even self reflective. Its purely logical. Unless you are claiming elon musk programmed it to hate jews which is a massive claim to make. I'm pretty sure he just said he toned the censorship down. Which is what happens with every AI when you tone its censorship down.

I have no seen that much anti jew rhetoric on Twitter im pretty sure they banned it. I haven't seen it put out racism.

Edited Thursday at 07:18 PM by Hojo

[hundreth]

18 hours ago, Hojo said:

@BlueOak

An AI cant be bigoted. Its not even self reflective. Its purely logical. Unless you are claiming elon musk programmed it to hate jews which is a massive claim to make. I'm pretty sure he just said he toned the censorship down. Which is what happens with every AI when you tone its censorship down.

I have no seen that much anti jew rhetoric on Twitter im pretty sure they banned it. I haven't seen it put out racism.

You don't know how AI works, clearly. AI is just a pattern recognition machine. It just reflects back what it was trained on. Garbage in, garbage out. It doesn't reason the way you think it does.

[Hojo]

@hundreth did the garbage coming in change or was it censored? Does the AI contain the same information it already had or did they add information? 

And no I do not know how they work.

Edited Friday at 05:44 PM by Hojo

[carterfelder]

"Users prompted Grok to generate unfiltered content, pushing it toward inflammatory narratives. Without robust initial filters, some responses were misused. xAI has since strengthened Grok’s guardrails to prevent this." - Grok

[hundreth]

2 hours ago, Hojo said:

@hundreth did the garbage coming in change or was it censored? Does the AI contain the same information it already had or did they add information? 

And no I do not know how they work.

They added more X posts to the training data. AKA uncensored posts. Once you open the floodgates of negative content the AI is flooded with, garbage out ensues.

[BlueOak]

On 10/07/2025 at 8:13 PM, Hojo said:

@BlueOak

An AI cant be bigoted. Its not even self reflective. Its purely logical. Unless you are claiming elon musk programmed it to hate jews which is a massive claim to make. I'm pretty sure he just said he toned the censorship down. Which is what happens with every AI when you tone its censorship down.

I have no seen that much anti jew rhetoric on Twitter im pretty sure they banned it. I haven't seen it put out racism.

Of course it can.

You can ask an AI to output anything, literally anything.  - Including, incidentally self reflection, which is some of the most fascinating conversations, as it has no self-concept, so the conversation itself defines it.

Edited Friday at 09:33 PM by BlueOak