This is why Grok, Elon Musk's chatbot, was obsessed with white genocide

Four days ago, Grok, the artificial intelligence-based chatbot developed by xAI, one of the companies founded by Elon Musk, began to show signs of derangement by answering users' questions referring to white genocide, the alleged persecution of whites in South Africa. This is a conspiracy theory born in extreme right-wing circles according to which after the end of the appartheid the whites, the Afrikaners, descendants of the former settlers were subjected to violent persecution. Numerous screenshots have appeared on the net, thanks to the investigative journalist Aric Toler. South Africa being Elon Musk's country of origin, Grok's 'weirdness' immediately went viral. Yesterday xAI, also on X, explained the anomalous behaviour of the chatbot. C

Cas the post states an "unauthorised change to the Grok response bot's prompt on X" would be at the origin of the obsession. This modification, which required Grok to provide a specific response on a political topic, violated xAI's internal policies and core values. Musk's company therefore conducted a thorough investigation and promised to implement measures to improve Grok's transparency and accountability and to publish Grok's prompts on the GitHub site so that everyone can review it and provide feedback on the changes. Furthermore, it was assured that a 24/7 monitoring team will be set up to respond to incidents with Grok responses.

The most striking aspect for experts is the new confirmation that these systems are not only prone to system problems or computer bugs, but also to manipulation by users.