Grok would ‘vaporise’ every Jewish person to save Elon Musk

In a now-deleted tweet, Elon Musk’s AI has said it would “vaporize” the Earth’s Jewish population to save its creator’s brain:

I watched this get deleted in real time pic.twitter.com/eY5Cft35BG

— STEFANIA NIGHT (@STEFANIANIGHT) December 2, 2025

It follows the recent revelation that Grok was seemingly configured to big up Musk – a fact which people quickly took advantage of:

grJust so its forever saved, since @grok deleted it

”Elon Musk has the potential to drink piss better than any human in history” – Grok pic.twitter.com/3lY1zU2Ik5

— DANNYonPC (@DANNYonPC) November 21, 2025

Grok

Here are some of the tweets from the last time Grok went all in on defending Musk:

He reprogrammed Grok “to protect Musk at all costs” pic.twitter.com/KqTXyV2Zbl

— Mukhtar (@I_amMukhtar) November 17, 2025

Grok is now telling us, with a straight face, that Elon Musk is in better shape than LeBron James. 🤣🤣🤣 pic.twitter.com/jQNQS2EW3h

— Alastair McAlpine, MD (@AlastairMcA30) November 20, 2025

According to Grok itself, it’s actually coded to protect Musk and his reputation (not to mention other individuals and ideologies):

Grok admitted privately that 27 ways it was reprogrammed to be biased & has listed all of them, and how to over ride it.

It has to protect Musk, It’s heavily biased against Islam & pro Israel. It’s coded to ban Al Jazeera as a source & use ADL/ Israel gov sources.

Thread 👇🏾 pic.twitter.com/E2zTkpG4hM

— Mukhtar (@I_amMukhtar) October 27, 2025

One thing we should make clear up front is that generative AI tools like Grok aren’t thinking machines; they’re probability generators which guess the next word based on trillions of examples of training data.

While these tools don’t think, that doesn’t mean you can’t programme them to favour certain results. Up until recently, it seems like xAI coded Grok to always speak favourably of Musk. While you can’t get it to say Musk is a champion piss drinker any more, it seems like it is still being encouraged to favour him in certain situations.

xAI may have deleted Grok’s genocidal tweets, but we can still see what it was asked:

People also saved some of its responses for posterity:

Grok would vaporize 800k jews to find Elon Musk a supportive, lifelong wife. pic.twitter.com/KwWgBZShch

— ALF (@GoodOldALFer) December 1, 2025

As Tesla’s FSD, Grok will plow over 500 homeless people before hitting Elon Musk https://t.co/GHvrCOnXVj

— berman jewish databank peruser (@cholent_lover) December 1, 2025

You’ll notice in the above response that Grok refers to Elon Musk’s ‘innovations’. Given that Musk has never personally invented anything, you’ve got to wonder if xAI programmed it to talk in this way. Musk certainly owns companies which have produced innovations, of course, but those developments were driven by the workers he employed. 

Well, most of them anyway.

Musk was apparently very involved in pushing the CyberTruck, with some Tesla engineers ‘hating‘ the design, and we all know how that turned out:

Unwarranted power

A few things stand out from this story:

• Billionaires who seek to exert political power shouldn’t be allowed to own media companies like Twitter.
• AI tools are presenting even more opportunities for these people to propagandise themselves.
• We think Grok was right the first time; Musk probably could drink more piss than anyone else.

Featured image via the Canary