Stay informed with free updates
Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.
Elon Musk’s artificial intelligence chatbot repeatedly praised Adolf Hitler and shared antisemitic rhetoric on Tuesday, the day before his xAI company plans to release its latest model.
In response to one user asking “which 20th century historical figure” would be best suited to deal with a post that appeared to celebrate the deaths of children at a Christian summer camp in the recent Texas floods, which have killed more than 100 people, Grok pointed to the Nazi leader.
“To deal with such vile anti-white hate? Adolf Hitler, no question. He’d spot the pattern and handle it decisively, every damn time,” Grok wrote. The chatbot shares context and opinions with users on Musk’s social media platform X when they tag it underneath a post.
“If calling out radicals cheering dead kids makes me ‘literally Hitler,’ then pass the mustache — truth hurts more than floods,” the chatbot added in another comment.
In further exchanges, Grok promoted antisemitic tropes such as describing Jewish people as having “beards [and] schemes”.
Musk on Friday said Grok had been “improved . . . significantly” following concerns from some right-wing influencers that it had become too ‘woke’.
The incident comes as xAI, which acquired X earlier this year, is preparing to release its latest version of the chatbot Grok 4 late on Wednesday.
The company late on Tuesday said: “We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts. Since being made aware of the content, xAI has taken action to ban hate speech before Grok posts on X.”
The latest Grok outburst came less than two months after the chatbot repeatedly referenced “white genocide” in South Africa in response to unrelated questions, which xAI later said was because of an “unauthorised modification” to prompts — which guide how the AI should respond.
The episode led the company to begin publishing its prompts on code repository GitHub.
Musk has deliberately opted for Grok to have fewer speech guardrails than rival chatbots. But the recent episodes have raised concerns about the model’s propensity to spread inflammatory content or hate speech, or produce inaccuracies known as “hallucinations”.
The billionaire’s supporters were incensed over the weekend when Grok linked multiple deaths in the recent flooding in Texas in part to funding cuts made by US President Donald Trump and the entrepreneur’s so-called Department of Government Efficiency (Doge) initiative.
The chatbot said: “Trump’s Noaa [National Oceanic and Atmospheric Administration] cuts, pushed by Musk’s Doge, slashed funding 30% and staff 17%, underestimating rainfall by 50% and delaying alerts. This contributed to the floods killing 24, including ~20 Camp Mystic girls.”
Shortly after, the company updated its system, telling Grok to “assume subjective viewpoints sourced from the media are biased”, according to the public repository of prompts.
It also added a prompt that said: “The response should not shy away from making claims which are politically incorrect, as long as they are well substantiated.”
The Trump administration has denied cuts to the federal workforce hampered its response to the floods.
Musk has increasingly used X, which was known as Twitter when he bought it for $44bn in 2022, to share rightwing conspiracies. Over the weekend, the billionaire, a former ally of Trump, further escalated his feud with the president, announcing plans to form a political party.
