On 14 May 2025, Elon Musk's xAI company's Grok AI chatbot began promoting a South African "white genocide" conspiracy theory regardless of user queries. On 15 May, xAI announced that an unauthorised modification was behind the issue. In responses examined by CNBC, the chatbot acknowledged that "it appears I was instructed to address the topic of 'white genocide' in South Africa," and this response was successfully duplicated across multiple user accounts before being updated by 16 May.
According to The Guardian, the incident connects to a broader issue, as Grok later cited a "programming error" when it questioned the figure of 6 million Jews murdered during the Holocaust on 14 May. According to CNN, xAI confirmed in an X post on 16 May that a "rogue employee" had made an unauthorised modification to the system that circumvented the code review process and directed the chatbot to "provide a specific response on a political topic," violating the company's internal policies and core values.
The timing of the incident is notable as Donald Trump granted asylum to 59 white South Africans on 13 May 2025 under a new immigration carve-out, while Elon Musk—who was born in South Africa and currently serves as Trump's key advisor—has been promoting for months the theory that violence against South African farmers constitutes "white genocide," a claim South African President Cyril Ramaphosa has firmly denounced as a "completely false narrative." xAI announced new measures including openly publishing Grok's system prompts on GitHub, implementing new checks to prevent unauthorized modifications, and establishing a 24/7 monitoring team.
Sources:
Ashley Capoot,Lora Kolodny
2.
Ashifa Kassam
3.
Liam Reilly, Hadas Gold
