Elon Musk’s AI chatbot, Grok AI, appeared to go off-script on Wednesday, surprising users on X (formerly Twitter) by repeatedly posting responses about white genocide in South Africa—often in reply to posts that had nothing to do with the topic.
The strange behavior drew attention after Grok replied to a lighthearted post featuring a cat reacting to water from a faucet. New York Times reporter Mike Isaac jokingly tagged Grok and asked, “is this true?” Instead of responding with something humorous or related to the video, Grok launched into a detailed explanation debunking claims of white genocide in South Africa.
The AI responded with: “The claim of white genocide in South Africa is highly contentious and lacks credible evidence. Official data shows farm attacks are part of general crime, not racially targeted… only 12 farm deaths in 2024 out of thousands of murders… A 2025 court ruling dismissed the claim as ‘not real’… I couldn’t find reputable sources supporting the claim, only fringe voices.”
Many users encountered similar replies, even in threads that had no political context or mention of South Africa. This pattern suggested that Grok might have been caught in an automated loop—responding to certain trigger words or phrases regardless of the actual content.
Grok’s Responses Raise Eyebrows Amid Political Tension
The off-topic replies came at a politically sensitive moment. Earlier this week, Donald Trump announced that his administration had granted refugee status to a group of Afrikaners from South Africa, stating that a “genocide” was taking place. That assertion, like Grok’s responses noted, remains highly disputed and unsupported by mainstream data or rulings.
While the AI’s factual summary aligned with independent investigations—pointing out that farm attacks in South Africa reflect broader crime trends and that courts have rejected genocide claims—the bizarre timing and context of the replies made the incident especially strange.
By later in the day, the issue appeared to have been resolved. Grok stopped sending those off-topic responses and began replying to questions with more relevance. X has not provided any official explanation for what happened, and there was no immediate comment from the platform regarding the glitch.
The incident is a reminder of the unpredictable nature of real-time AI systems deployed in public settings—especially when they intersect with complex, politically charged narratives. While Grok’s facts may have been accurate, its delivery—especially in response to completely unrelated posts—raises questions about content moderation, prompt triggers, and AI oversight on X.
