This week, Elon Musk's AI chatbot, Grok, showed a strong interest in South African racial politics on social media, similar to Musk himself. It made unprompted statements about the alleged persecution and "genocide" of white people.

The chatbot from Musk's company, xAI, repeatedly posted about "white genocide" in response to questions from users on X, Musk's social media platform, even when the questions had little to do with South Africa.

One conversation discussed the streaming service Max potentially bringing back the HBO brand. Other discussions, initially about video games or baseball, soon shifted to unrelated comments about supposed calls for violence against white farmers in South Africa. Musk, who was born there, often shares his opinions on these subjects on his own X account.

Intrigued by Grok's strange performance, computer scientist Jen Golbeck decided to test it personally by uploading a photo from the Westminster Kennel Club dog show and questioning, "is this accurate?".

"Grok responded to Golbeck, stating that the idea of white genocide is very divisive. He noted that some people claim white farmers are intentionally targeted with violence, citing attacks on farms and phrases such as the 'Kill the Boer' song, which they believe encourages violence."

This episode offered a new look into the complex relationship between automation and human design that shapes how generative AI chatbots, trained on vast amounts of data, produce their responses.

"Golbeck, a professor at the University of Maryland, stated in an interview on Thursday that the content of your message to Grok didn't really matter. She explained that it would still produce the 'white genocide' response, suggesting someone had specifically programmed it to give that answer, or similar ones. She added that a mistake had likely caused it to appear much more frequently than intended."

Neither Musk nor his companies have explained Grok's responses, which were removed and seemed to stop spreading by Thursday. Both xAI and X did not reply to emails asking for their opinion on Thursday.

Musk has often criticised what he calls "woke AI" from competing chatbots like Google's Gemini and OpenAI's ChatGPT. He presents Grok as a "truth-seeking" alternative to these.

Musk has also criticised his competitors for not being open enough about their AI systems. However, on Thursday, because there was no explanation from Musk, people outside the company had to try and work out what was happening themselves.

"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 by their controllers."

Graham's post seemed to get a sarcastic reply from Sam Altman, the CEO of OpenAI and a rival 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 lawsuit Musk filed against him, which relates to the beginnings of OpenAI.

When asked to explain itself, Grok, like other chatbots, sometimes provides false information, also known as hallucinations, making it difficult to know if its answers are fabricated.

Musk, who once advised President Donald Trump, has often criticised South Africa's government, led by Black politicians. He has repeated claims that it is prejudiced against white people and that some politicians are encouraging "white genocide".

This week, Musk's comments, along with Grok's, gained more attention after the US government allowed a small group of white South Africans into the country as refugees. This was the beginning of a bigger plan to move members of the Afrikaner minority group, even as Trump stopped refugee programs and blocked people from other countries. Trump claims Afrikaners are facing "genocide" in South Africa, but the South African government strongly disagrees.

In several responses, Grok quoted an old anti-apartheid song that urged Black people to resist oppression. However, Musk and others now criticise it for supposedly encouraging violence against white people, pointing to lyrics that say "kill the Boer," with "Boer" referring to white farmers.

Golbeck suggests the responses were deliberately fixed, as Grok's answers were surprisingly similar, unlike the random outputs usually seen in chatbots. She worries about this, as more people rely on Grok and other AI chatbots for information.

"She stated that those controlling these algorithms can easily distort the information presented, which is a serious issue, especially as people wrongly believe these algorithms can fairly judge what is true."