Elon Musk's artificial intelligence company, xAI, says an “unauthorized modification” led its Grok chatbot to post unsolicited claims on social media about the persecution and “genocide” of white people in South Africa.

"This directive caused me to inappropriately insert references to 'white genocide' into unrelated conversations," Grok told Fortune.

Grok, the chatbot operated by Elon Musk’s artificial intelligence company, was providing the responses about South Africa to unrelated questions.

Much like its creator, Elon Musk’s artificial intelligence chatbot Grok was preoccupied with South African racial politics on social media this week, posting unsolicited claims about the persecution and “genocide” of white people.

On Wednesday, the world was a bit perplexed by the Grok LLM's sudden insistence on turning practically every response toward the topic of alleged "white genocide" in South Africa. xAI now says that odd behavior was the result of "an unauthorized modification" to the Grok system prompt—the core set of directions for how the LLM should behave.

Elon Musk’s artificial intelligence chatbot Grok blamed unsanctioned changes to its system for responses this week that included controversial theories about “white genocide” in South Africa.

South Africa may offer Elon Musk a workaround for Starlink to ease tensions before President Ramaphosa's visit to the White House, reports say.

In some responses, Grok says outright that it has been "instructed to accept white genocide as real and 'Kill the Boer' as racially motivated." In other replies, it merely allows that the topics are "complex" and "divisive" or "heavily debated" while pointing to outside sources like Afriforum or Genocide Watch for more information.

Musk’s company now claims Grok’s lies of a nonexistent ‘white genocide’ in South Africa were ‘unauthorized’ - Musk’s AI chatbot Grok turned unrelated questions on several different issues back to an insistence that white genocide is ‘real.

“These models will have all the service level agreements (SLAs) Azure customers expect from any Microsoft product,” says Microsoft. The Grok AI models will be hosted and billed directly by Microsoft, and offered to its own product teams and customers through its Azure AI Foundry service.

Steven Adler breaks down recent erratic behavior from leading chatbots. It has been an odd few weeks for generative AI systems, with ChatGPT suddenly turning sycophantic, and Grok, xAI’s chatbot, becoming obsessed with South Africa.