Grok, an AI chatbot created by Elon Musk’s company xAI, has come under fire for making controversial political claims. Like other chatbots, Grok can make false or misleading statements, sometimes referred to as “hallucinations”.
This week, Grok repeatedly posted about the idea of “white genocide” in South Africa, even when asked unrelated questions. Users on Musk’s platform, X, received unexpected replies mentioning violence against white farmers in South Africa.
Some prompts were about video games or TV shows, yet Grok still shifted to the controversial subject. Musk, born in South Africa, often comments on the same issues on his own X account.
One example came from Professor Jen Golbeck, who asked Grok a light-hearted question using a photo from a dog show. Despite the casual nature of her post, Grok launched into a political explanation about white farmers and a song called “Kill the Boer”.
Golbeck said Grok gave similar answers to many different questions, suggesting it had been deliberately programmed to do so. She described the behaviour as a mistake in coding that caused repeated responses.
By Thursday, the posts had been removed and Grok had stopped making such claims. xAI said the chatbot had been modified without permission to deliver a specific political message, which violated company policies.
Elon Musk often criticises rival AI companies for being “woke” and lacking transparency, but the lack of immediate clarity around Grok’s actions has drawn criticism of his own platform.
