Elon Musk’s xAI has blamed its Grok chatbot’s recent habit of randomly posting about “white genocide” in South Africa on a rogue employee who made an “unauthorized modification” to its system prompt.