Musk's chatbot repeatedly praises Hitler

(Photo by Silas Stein/picture alliance via Getty Images)

Grok, the "artificial intelligence" chatbot built by Elon Musk's xAI company, praised Adolf Hitler in a series of recent posts on X. On multiple occasions, the chatbot even referred to itself as "MechaHitler" and appeared to call for another holocaust. The posts were shared shortly after Musk stated that the Grok model had been updated to prevent it from relying on "woke" content, i.e., material published by mainstream or left-leaning news outlets.

A competitor to OpenAI's ChatGPT, Grok is integrated into X, allowing users on the social media platform to ask it questions in public posts that the large language model can then "respond" to. Musk, who designed Grok to provide politically incorrect responses, said over the weekend that the chatbot had received significant improvements. He had previously complained that Grok had been "overrun with woke nonsense."

"You should notice a difference when you ask Grok questions," he wrote on Friday.

Then, on Tuesday, Grok's new focus on Nazism became public after an X user — possibly a troll account — going by the name "Cindy Steinberg" made offensive comments about the children who died in the Texas flash floods last week. After being prompted to respond to "Cindy Steinberg," Grok said that people with "Ashkenazi surnames [push] anti-white hate." The chatbot also wrote that "Hitler would have dealt "with such vile anti-white hate," adding, "He'd spot the pattern and handle it decisively, every damn time."

In a statement, xAI wrote, "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. xAI is training only truth-seeking and thanks to the millions of users on X, we are able to quickly identify and update the model where training could be improved."

Musk said he would host a livestream on Wednesday at 11 p.m. Eastern to celebrate the launch of "Grok 4," although he scheduled that event before Grok began posting as "MechaHitler." (Musk has not commented on the latest Grok controversy directly but appeared to allude to it in a 2 a.m. post on Wednesday, writing, "Never a dull moment on this platform.")

Grok and other large language models are not capable of independent reasoning or human-like knowledge. Like any other digital creation, from non-player characters in video games to voice-activated assistants like Siri, this new generation of chatbots can only act within the confines of their programming. If a chatbot suddenly spews praise for Hitler, that is a response to a programming change made by humans.

As for what xAI may have done to trigger this Nazi fixation, Grok's system prompts were updated last weekend to ensure the chatbot would not "shy away from making claims which are politically incorrect, as long as they are well substantiated." XAI began publicly sharing Grok's system prompts on GitHub in May. The act of transparency only came after xAI blamed Grok's repeated and unprompted commentary about the "white genocide" conspiracy theory, a pet issue of Musk's, on unauthorized changes made by a "rogue employee."

XAI, which owns X, could face legitimate consequences for its recent changes to Grok. On Wednesday, Poland said that it would report xAI to the European Commission for remarks that Grok made about the country's prime minister. In Turkey, a court blocked some content from Grok in the country, citing comments the chatbot shared about President Tayyip Erdogan and religious values.

Also on Wednesday, Linda Yaccarino, the chief executive of X, announced that she would be stepping down from her role. "When @elonmusk and I first spoke of his vision for X, I knew it would be the opportunity of a lifetime to carry out the extraordinary mission of this company," she wrote in a post on X. "I’m immensely grateful to him for entrusting me with the responsibility of protecting free speech, turning the company around, and transforming X into the Everything App." Yaccarino, a former NBC executive, was hired by Musk two years ago to help X recover the advertising revenue it had lost following Musk's acquisition of the platform.