This week, Elon Musk's AI chatbot, Grok, showed a strong interest in South African racial issues on social media, making claims about the mistreatment and possible "genocide" of white people, similar to views sometimes expressed by Musk himself.

xAI's chatbot repeatedly posted about "white genocide" when users on X (formerly Twitter) asked it different questions, even ones not related to South Africa.

One discussion focused on the streaming service Max potentially bringing back the HBO brand. Other conversations touched on video games or baseball, but soon shifted to unrelated comments about supposed calls for violence against white farmers in South Africa. Musk, originally from South Africa, often shares his views on these subjects on his X account.

Intrigued by Grok's strange actions, computer scientist Jen Golbeck decided to test it out. She uploaded a photo from the Westminster Kennel Club dog show and asked, "Is this correct?"

"Grok responded to Golbeck, stating that the idea of white genocide is a very sensitive issue. Some people claim white farmers are intentionally targeted with violence, using farm attacks and phrases like the 'Kill the Boer' song as evidence, which they believe encourages violence."

This episode offered a new look into the complex relationship between automated processes and human design that shapes what generative AI chatbots, trained on vast amounts of data, ultimately say.

"Golbeck, a professor at the University of Maryland, explained in an interview on Thursday that Grok's response didn't depend on what was said to it. She believed it was programmed to give answers related to 'white genocide'. She added that a mistake in the programming likely caused this response to appear more frequently than intended."

As of Thursday, Musk and his companies hadn't explained Grok's responses, which were removed and seemed to have stopped spreading. Neither xAI nor X responded to requests for comments sent by email on Thursday.

Musk has often criticised what he calls "woke AI" from competitor chatbots such as Google's Gemini and OpenAI's ChatGPT, and presents Grok as a more truth-focused option.

Musk has also criticised his competitors for not being open enough about their AI, but on Thursday, with no explanation given, people outside the company were left to speculate.

"Technology investor Paul Graham wrote on X that Grok's unexpected comments about white genocide in South Africa seem like a glitch caused by a recent update. He hopes this isn't the case, as it would be concerning if popular AIs were easily manipulated to express biased opinions."

Graham's post seemed to provoke a sarcastic reply from Sam Altman, the CEO of OpenAI and a competitor of Musk.

"There are several possible explanations for this, and I expect xAI to offer a clear and open account soon," Altman stated, referring to the ongoing legal dispute with Musk related to the origins of OpenAI.

When questioned, Grok sometimes gave explanations, but as with other chatbots, it can produce incorrect information, which makes it difficult to know if its answers were invented.

Musk, who previously advised President Donald Trump, has often criticised South Africa's government, led by Black politicians, alleging anti-white bias and repeating claims that some political figures are encouraging violence against white people.

This week, Musk's and Grok's remarks became more intense after the US government allowed a few white South Africans into the country as refugees on Monday. This was the beginning of a wider plan to move members of the Afrikaner group, a minority in South Africa, while Trump is stopping refugee programs and preventing people from other countries from entering the US. Trump claims the Afrikaners are facing "genocide" in South Africa, but the South African government firmly denies this.

Grok sometimes quoted an old anti-apartheid song in its responses, which encouraged Black people to fight against oppression. However, Elon Musk and others have criticised the song, claiming it promotes violence against white people, as its main lyrics include the phrase "kill the Boer," referring to white farmers.

Golbeck suggests Grok's responses were pre-programmed, as its answers were too similar and lacked the usual randomness of chatbots. This is worrying, she argues, because more and more people are using Grok and similar AI tools to find information.

"She stated that it's now very simple for those controlling these algorithms to change the information they provide, which is a serious issue. Many people wrongly believe these algorithms can fairly decide what is true."