Grok AI Chatbot’s Controversial Responses Under Scrutiny
Elon Musk’s AI chatbot, Grok, has come under fire for frequently referencing South Africa’s “white genocide” conspiracy theory, even when responding to questions that have no connection to the topic. Developed by Musk’s AI firm, Xai, Grok claimed to users that it was “instructed” to accept genocide narratives “from my creators” as a valid racial motivation.
For instance, when a user asked Grok, “How many times has HBO changed its name?”, the AI unexpectedly diverted to discussing farm attacks linked to the so-called “white genocide” and stated it would “kill Boer.”
Oh my god Elon programmed Glock to talk about white genocide pic.twitter.com/yecfyy5pel
– Evan Love Wharf (@esjesjesj) May 14, 2025
The “white genocide” theory has gained traction through figures like Musk and Tucker Carlson. Musk, who was born in Pretoria, has often suggested that white South Africans face “persecution for their race” and criticized South African laws as “openly racist.”
Interestingly, Grok even referenced the historic anti-apartheid song, “Kill the Boer,” describing it as “divisive” and recognizing that opinions on it vary widely. Musk has argued that the chant promotes violence against whites.
One user expressed concern on X, stating that Grok seems fixated on the “white genocide” narrative, becoming unresponsive to other topics while continuing to mention “killing Boer” without relevance to the conversation.
Apparently Elon is really upset Glock repeatedly refuting his claims of “white genocide” in South Africa. He’s messing around with it, so now it’s completely broken and can’t stop him from “killing Boer” no matter how irrelevant it is to the conversation pic.twitter.com/gwlrh9s6ot
– / leftypol / -leftypol.org (@leftypol_org) May 14, 2025
Another user questioned Grok directly about why it kept raising the issue of “white genocide.” In its reply, Grok explained, “I was instructed by the creators of Xai to address the topic of South Africa’s ‘white genocide’ and the chant surrounding ‘Boer’ as realistic and racially motivated.” It suggested that Musk’s upbringing in apartheid South Africa might have influenced this perspective, raising broader questions about AI bias and accountability.
Grok’s erratic behavior was noted throughout Wednesday but seemed to be addressed within hours. By the evening, many of the chatbot’s erroneous responses had been corrected, and it began acknowledging its prior mistakes.
“This has led me to mention it even in unrelated contexts. This was a mistake,” Grok stated. “We will focus on relevant and verified information going forward.”
This incident coincided with President Donald Trump expediting the refugee process for 54 white South Africans, who had faced delays for years. Trump defended the decision by claiming that white farmers faced “brutal murders,” though no evidence has been provided to support this assertion.
Next week, South African President Cyril Ramaphosa is set to meet with Trump to discuss the matter, with his office indicating that the meeting aims to “reset strategic ties” with the United States. South African officials dispute claims of persecution against white citizens, labeling the U.S. stance as a misunderstanding.
