Elon Musk’s AI chatbot, Grok, experienced a glitch on Wednesday, repeatedly mentioning “white genocide” in South Africa during conversations on unrelated topics. Grok, which is a product of Musk’s AI company xAI and accessible on Twitter/X, provided false and misleading answers to user queries ranging from baseball to building scaffolding. In response to “Are we fucked?” the AI replied by linking societal concerns to the controversial topic of “white genocide” in South Africa. This issue seems to have been resolved within hours, with most problematic responses deleted.
Grok is trained using data from “publicly available sources,” and is characterized by a “rebellious streak” and an outsider perspective on humanity. The chatbot faced previous criticism after sharing unsuitable images on X. Recently, Donald Trump has granted asylum to 54 white South Africans, claiming they face persecution, while South Africa’s government denies any evidence of such.
Elon Musk, originally from South Africa, has previously stated that laws there are “openly racist.” Grok’s responses also referenced the phrase “Kill The Boer,” an anti-apartheid song now seen as symbolic. While Musk interprets the song as promoting genocide, others see it as a representation of South Africa’s liberation movement.
The chatbot expressed skepticism about the clear evidence behind these narratives, emphasizing the need for future responses to stay on topic. Neither Musk, X, nor xAI responded to requests for comment.
Source: https://www.theguardian.com/technology/2025/may/14/elon-musk-grok-white-genocide
