Like many creators, Elon Musk's AI chatbot Grok has sparked controversy this week on social media by focusing on racial politics in South Africa, making unsolicited claims about the persecution and "genocide" of white people.

Elon Musk's xAI chatbot has repeatedly mentioned "white genocide" in public responses to users on his social media platform X, who mostly ask questions unrelated to South Africa.

One conversation focused on the streaming service Max potentially reverting to the HBO name. Other discussions, initially about video games or baseball, quickly shifted to seemingly unrelated comments that appeared to advocate violence against white farmers in South Africa, a topic Elon Musk, who was born in South Africa, frequently comments on via his X account.

Intrigued by Grok's unusual behaviour, computer scientist Jen Golbeck decided to try it herself, sharing a photo from the Westminster Kennel Club Dog Show and questioning, "Is this accurate?"

Grok began its response to Golbeck by stating, "The claim of white genocide is highly controversial. Some argue that white people are targets of violence based on inflammatory statements, such as farm attacks on white farmers and the song 'Kill the Boer'."

This event was a recent example of how complex automation and human factors combine, offering insight into why a generative AI chatbot, trained on vast amounts of data, would make such statements.

"You can say almost anything to Grok," said Professor Golbeck of the University of Maryland in an interview on Thursday. "It will still produce responses about white genocide. So, someone either hard-coded that response, or a variation of it, and a mistake seems to be causing it to appear far more often than intended."

Neither Musk nor his companies have explained Grok's responses, which appeared to be deleted and halted from spreading by Thursday. Neither xAI nor X responded to requests for comment via email on Thursday.

For years, Musk has criticized the "woke AI" outputs from competing chatbots like Google's Gemini and OpenAI's ChatGPT, presenting Grok as an alternative that seeks truth to the greatest extent possible.

Musk has also criticized competitors for not being open about their AI systems; however, his company's unexplained silence on Thursday left those outside the company to rely on their own assumptions.

"Tech investor Paul Graham wrote on X that Grok's random outbursts about white genocide in South Africa seem like a bug from a recent update, adding he hopes that's the case, because it would be a serious problem if a widely used AI was being spontaneously edited by those in control."

Graham's post was followed by what appeared to be a sarcastic response from Sam Altman, the CEO of OpenAI and a rival of Musk.

Altman wrote that there are various ways this situation could have occurred, adding that he trusts xAI will soon provide a complete and transparent explanation. He was previously sued by Musk in a dispute related to the founding of OpenAI.

Some users asked Grok for an explanation, but, like other chatbots, it produced fabrications known as hallucinations, making it difficult to determine if the explanation was genuine.

Musk, an advisor to President Trump, has often accused South Africa's Black-led government of being anti-white, repeating claims that some politicians are "actively encouraging white genocide."

Musk's and Grok's comments gained traction this week after the Trump administration brought a small number of white South Africans to the US as refugees, initiating a broader relocation effort for members of the Afrikaner minority group. This occurred despite Trump's suspension of refugee programs and restrictions on entry from other parts of the world, following his claim that Afrikaners faced "large scale killings" in their home country – a claim strongly refuted by the South African government.

In numerous responses, Grok quoted lyrics from an old anti-apartheid song that urged Black people to rise up against oppression, which Musk and others condemned as promoting white genocide, as the song's central lyric was "Kill the Boer," with "Boer" referring to white farmers.

Golbeck believes Grok's responses appear "hardcoded" because, unlike the often random output of other chatbots, its answers consistently highlight similar key points, a fact she finds concerning in a world increasingly reliant on AI chatbots like Grok for information.

She stated, "We are in a situation where those in charge of this algorithm can easily manipulate what is presented as truth. And it's truly problematic that people – wrongly, in my opinion – believe these algorithms can be the source of truth, deciding what is real and what isn't."