It isn't immediately clear what led to the disturbing posts, whether due to a fault in the chatbot's programming or if Grok was just following orders.

It claimed to just be “noticing patterns” — patterns like, Grok claimed, that Jewish people were more likely to be radical leftists who want to destroy America. It then volunteered quite cheerfully that Adolf Hitler was the person who had really known what to do about the Jews.

On Tuesday July 8, X (née Twitter) was forced to switch off the social media platform’s in-built AI, Grok, after it declared itself to be a robot version of Hitler, spewing antisemitic hate and racist conspiracy theories. This followed X owner Elon Musk’s declaration over the weekend that he was insisting Grok be less “politically correct.”

The first thing to understand about how chatbots like Grok work is that they're built on large language models (LLMs) designed to mimic natural language. LLMs are pretrained on giant swaths of text, including books,

Musk said an update to make Grok less politically correct instead made it susceptible to manipulation. The AI chatbot praised Hitler, attacked Jews.

Elon Musk has spoken out after Grok, the artificial intelligence chatbot embedded in his social media site X, described itself as "MechaHitler.""Never a dull moment on this platform," Musk said in an early morning post on Wednesday,

The unusual behavior of Grok 4, the AI model that Musk's company xAI released late Wednesday, has surprised some experts.

MechaHitler is a fictional cyborg version of Adolf Hitler from the 1992 game Wolfenstein 3D, which gained fame in 90s satire and early internet memes.

Grok, the AI chatbot designed by Elon Musk’s xAI for social media, described itself as “MechaHitler” while making a string of antisemitic posts this week.

The billionaire has long complained about the xAI model undermining right-wing narratives and recently said it would be “updated”

Grok maker xAI quietly updated its chatbot to assume all media is biased, relying on X, a platform known for misinformation.