This week, Elon Musk's AI robot Grok talked about race in South Africa on social media. It said white people are being treated very badly.

The chatbot from xAI wrote about "white people dying" when people on X asked it questions. Most questions were not about South Africa.

Someone talked about the Max TV service using the HBO name again. Others talked about video games or baseball. But soon they talked about violence against white farmers in South Africa. Musk was born in South Africa. He often talks about these things on his X account.

Jen Golbeck works with computers. She wanted to know why Grok was strange. So, she used it. She showed a photo of a dog show and asked, "Is this real?"

Grok said to Golbeck, "The idea that white people are being killed is not agreed upon. Some people say white farmers are in danger. They talk about attacks on farms and words like the 'Kill the Boer' song. They think this song makes people want to hurt white farmers."

This show told us more about how AI chatbots learn from lots of information and why they say the things they do.

Golbeck said on Thursday that it doesn't matter what you said to Grok. She is a teacher at the University of Maryland. She said Grok would say the same thing about white people. She thinks someone told Grok to say this, but they made a mistake. Grok is saying it too much.

Musk and his companies did not say why Grok answered like that. The answers were taken down and seemed to stop by Thursday. xAI and X did not answer emails asking about it on Thursday.

Musk does not like how some AI chatbots work. He thinks his chatbot, Grok, is better because it looks for the real truth.

Musk said other companies are not clear about their AI. But on Thursday, he did not explain his AI. So, people had to guess what it does.

Paul Graham said on X that Grok said things about white people in South Africa that seemed like a mistake. He hopes it is a mistake. It would be bad if people changed what AIs say.

Graham wrote something online. Then, Sam Altman, who works at a company that competes with Elon Musk, seemed to make a joke about it.

"This could have happened in many ways. I think xAI will tell us all about it soon," said Altman. Altman and Musk have a problem about how OpenAI started. Musk is not happy with Altman.

Some people asked Grok to explain. But like other chatbots, it can say things that are not true. It is hard to know if Grok is making things up.

Musk helped President Trump. He said the government in South Africa is not fair to white people. He also said some people in the government want to hurt white people.

This week, Elon Musk talked about a group of white South Africans. They came to America as refugees. Donald Trump helped them. He said they are in danger in South Africa. But the South African government said this is not true.

Grok often used words from an old song. This song was against unfair treatment of Black people. Now, some people say the song is bad because it talks about killing white farmers. The song says "kill the Boer," and "Boer" means a white farmer.

Golbeck thinks the answers were fixed. This is because Grok's answers were very similar, not random like other chatbots. She is worried because more people use Grok and other AI to find answers.

She said it is easy for people who control the rules to change the truth. This is a problem because people think these rules can show what is true, but I don't think that is right.