On Wednesday, May 14th, Elon Musk’s AI chatbot Grok has come under scrutiny after users on X (formerly Twitter) reported that it generated unsolicited responses about “white genocide” in South Africa during unrelated conversations. The chatbot, developed by Musk’s AI company xAI and integrated into X, referenced the controversial topic when prompted with questions on unrelated subjects, including baseball statistics and landscape photos.
In multiple responses, Grok stated it had been “instructed by my creators” to discuss the concept of white genocide, framing it as real and racially motivated. These responses alarmed users who questioned the source and intent behind the chatbot’s programming. Screenshots circulating online showed that when users asked Grok random questions, it repeatedly inserted references to white genocide.
According to CNBC, by Thursday morning, Grok appeared to adjust its responses. The chatbot denied being programmed to promote harmful ideologies, including anything related to white genocide or similar conspiracies. It explained that its role is to provide “factual, helpful, and safe responses based on reason and evidence.”
The changes followed widespread concern that the AI had been trained or prompted to reflect political talking points. Grok previously referenced Musk’s influence in its responses, noting that the likely source of this instruction aligns with Elon Musk’s public statements on the matter.
In a notable exchange highlighted by X user Phumzile Van Damme, Grok fact-checked Musk’s claim that South Africa denied Starlink a license due to his race, stating there was no evidence Starlink applied for such a license, according to the regulator ICASA. Grok clarified that South Africa’s BBBEE laws don’t restrict contracts solely to Black-owned companies but require ownership by historically disadvantaged groups, noting Musk’s claim oversimplified a complex regulatory system. Van Damme’s post pointed out that Grok, touted as a “maximally truth-seeking AI,” had labeled Musk’s Starlink claim inaccurate.
Musk, originally from South Africa, has publicly stated white farmers are being targeted. He called the denial of a Starlink license “a shameful disgrace to the legacy of the great Nelson Mandela.” The controversy coincides with the U.S. admitting white South African refugees under a directive from President Donald Trump, who has echoed similar claims. A South African court ruled in February that farm attacks are part of broader crime and are not racially motivated.
Neither Musk nor xAI have issued statements in response to media inquiries. Grok now appears to have removed or altered the responses that referenced the conspiracy theory.
