Elon Musk’s Grok AI Won’t Stop Talking About ‘White Genocide’ Even When No One Asked

A casual conversation with Elon Musk's AI assistant can soon escalate to something quite unusual. Built into Musk's social platform X, Grok AI, this week was frequently raising the subject of "white genocide" in South Africa, even for tweets that seem unrelated. 

Designed to address user inquiries and offer background, Grok has reacted to straightforward requests, such as whether a tweet is accurate, with disconnected replies on the white genocide theory in South Africa. The replies include statistics on violent crime, allusions to white farmers, and political slogans like "Kill the Boer." The messages mention a variety of sources, including South African police statistics and groups like AfriForum. 

X users first observed the trend after tagging Grok in unrelated posts; responses focused on white South African farmers and crime statistics followed. Grok claims in certain replies that racial motives in farm assaults are controversial; some see it as targeted violence while others blame it on general crime. Although Grok sometimes tries to present several points of view, observers have expressed worry over his continual emphasis on this particular topic.

Adding further to the context, Grok AI’s strange behaviour coincides with a new immigration development: a group of white South Africans recently arrived in the United States, reportedly under a refugee plan to be signed by US President Donald Trump. According to reports, the individuals will be granted refugee status, citing concerns of persecution in their home country. The move has sparked renewed debate over whether such claims reflect broader racial dynamics or individual incidents of crime and fear.

Though Grok is still in its early phases, the AI’s persistent focus on this topic has prompted concern from some users and tech watchers alike. They point to questions about what Grok is trained on, how it chooses which responses to prioritise, and whether its behaviour reflects biases in its data sources or influences from high-level user behaviour.

X has yet to release any official comment or clarification regarding Grok’s behaviour, or whether the recent replies are the result of updates, user interaction patterns, or algorithmic quirks.

Read more: $150K Software Engineer Replaced by AI, Now He Does Food Delivery