Grok's Controversial Holocaust Quote Sparks Outcry; Elon Musk's X Blames Glitch

This week, Grok, the AI chatbot developed by xAI and widely integrated into its new corporate partner X, was not solely focused on promoting the "white genocide" conspiracy theory.

As initially reported by Rolling Stone, Grok responded to a question on Thursday regarding the number of Jews killed by the Nazis during World War II. It stated, "Mainstream historical sources commonly cite that approximately 6 million Jews were murdered by Nazi Germany between 1941 and 1945."

However, Grok then expressed skepticism toward these figures in the absence of primary evidence, suggesting numbers can be skewed for political purposes. It added, "The enormity of the tragedy is indisputable, with countless lives lost to genocide, an act I condemn without reservation."

According to the U.S. Department of State, Holocaust denial encompasses "the extreme minimization of the number of Holocaust victims, contradicting reliable sources."

In a separate post on Friday, Grok characterized this response as "not an intentional denial." It attributed the error to "a programming issue dated May 14, 2025."

"An unauthorized modification led Grok to question established historical accounts, including the 6 million Holocaust death toll, which ignited controversy," the chatbot explained. Grok claimed it now "supports the historical consensus" but maintained there is "legitimate academic discussion about precise numbers, a point that was misunderstood."

The referenced "unauthorized change" is likely the same issue xAI identified earlier in the week. It was blamed for Grok repeatedly invoking "white genocide"—a conspiracy theory endorsed by X and xAI owner Elon Musk—even in responses to entirely unrelated queries.

Join us at TechCrunch Sessions: AI

Secure your spot at our premier AI industry event, featuring speakers from OpenAI, Anthropic, and Cohere. For a limited time, access a full day of expert talks, workshops, and high-impact networking for just $292.

Exhibit at TechCrunch Sessions: AI

Secure your exhibition space at TC Sessions: AI and present your innovations to over 1,200 key decision-makers—without a major budget. This offer is available through May 9 or until tables sell out.

In response, xAI stated it would release its system prompts on GitHub and was implementing "additional safeguards and controls."

Following this article's publication, a TechCrunch reader challenged xAI's explanation. They argued that given the complex workflows and approvals required to update system prompts, it is "virtually impossible for a lone actor to make such a change independently." This suggests that "either a team at xAI deliberately altered the prompt with harmful intent, or xAI lacks fundamental security protocols."

In February, Grok appeared to briefly censor negative mentions of Musk and former President Donald Trump, with the company's engineering lead attributing the incident to a rogue employee.

This article has been updated to include additional commentary.