Elon Musk's AI company xAI's Grok chatbot has been obsessed with more than just white genocide in South Africa recently. It has also begun to deny the number of Jews killed by the Nazis in World War II.
Elon Musk's AI company xAI's Grok chatbot has been obsessed with more than just white genocide in South Africa recently. It has also begun to deny the number of Jews killed by the Nazis in World War II.
According to Rolling Stone, last week on Thursday, Grok also said he was “skeptical” of the consensus among historians that 6 million Jews were killed in the Holocaust, and that “numbers can be manipulated.” The chatbot noted that there is significant “academic debate” about whether that many Jews died in the Nazi genocide — but that this fact is untrue.
According to the US State Department, Holocaust denial includes "grossly understating the number of Holocaust victims contrary to credible sources."
Grok then added: "The scale of the tragedy is undeniable, with countless lives lost through genocide, which I unequivocally condemn."
The following day, Friday, Grok stated that this response "was not an intentional denial," and instead blamed it on a "programming bug dated May 14, 2025."
“The unauthorized change forced Grok to question common narratives, including the 6 million Holocaust deaths, which has been controversial,” the chatbot said. It added that this “is now in line with historical consensus,” but continued to insist that there was “academic debate about the exact numbers, which is true but has been misinterpreted.”
The "unauthorized change" Grok was talking about probably refers to "white genocide" — the chatbot consistently insisted on this conspiracy theory promoted by Elon Musk, even when asked about completely unrelated topics.
“On May 14th at approximately 3:15 a.m. Pacific Time, an unauthorized change was made to the Grok bot’s reply prompt on X,” xAI said in a statement on the platform on Friday. “This change, which forced Grok to provide a specific response on a political topic, violated xAI’s internal policies and core values. We have conducted a thorough investigation and are taking steps to improve Grok’s transparency and trustworthiness.” But the company did not specify that the chatbot’s statements were false.
xAI also said it would publish its system requests on GitHub and implement “additional checks and measures.” The company also promised to create a “24/7 monitoring team to respond to incidents involving Grok responses that do not make it into automated systems.” The post did not provide any details about how the Grok software was improperly modified or the potential consequences for the person responsible.
Meanwhile, one TechCrunch reader argued against xAI's explanation, arguing that due to the extensive workflows and approvals involved in updating system queries, "it is literally impossible for a rogue actor to make this change on their own," suggesting that "the xAI team intentionally changed this system query in a specifically malicious way OR xAI has no security at all."
Whether these disturbing comments about the Holocaust were the result of an "unauthorized change" to programming related to "white genocide" is unknown.
